Home

DMS - DE Shaw Research

Contents

1. End 20 2 Licensed Companion Software 161 Desmond Users Guide Release 3 4 0 0 7 162 Chapter 20 Licenses and Third Party Software BIBLIOGRAPHY Bro 2004 C L Brooks HI A D MacKerell Jr M Feig Extending the treatment of backbone energetics in protein force fields limitations of gas phase quantum mechanics in reproducing protein conformational distributions in molecular dynamics simulations J Comput Chem 25 1400 1415 2004 Ben 1976 Charles H Bennett Efficient estimation of free energy differences from Monte Carlo data J Comp Phys 22 245 268 1976 Ber 1984 H J C Berendsen J P M Postma W F van Gunsteren A Dinola and J R Haak Molecular dynamics with coupling to an external bath J Chem Phys 81 3684 3690 October 1984 Don 2005 A G Donchev V D Ozrin M V Subbotin O V Tarasov and V I Tarasov A quantum mechanical polarizable force field for biomolecular interactions Proceedings of the National Academy of Science 102 7829 7834 May 2005 Ess 1995 U Essman L Perera M L Berkowitz T Darden H Lee and L G Pedersen A smooth particle mesh Ewald method J Chem Phys 103 19 8577 8593 1995 Fel 1995 S E Feller Y Zhang R W
2. where a Rtap Reut Reap and t x 1 x 1 22 or t x 1 x 1 3x 62 correspond to clswitch or c2switch respectively The average dispersion v is used to calculate energy and virial corrections due to cutoff in the van der Waals inter actions whenever such interactions are present in the force field and used by the selected nonbonded type If omitted Desmond calculates v based on the van der Waals terms and the atom types in the system 7 4 Nonbonded far interactions The nonbonded far electrostatic forces are configured as shown in force nonbonded far type gse pme nok ke ky kz transform c2c r2c r2c_2round auto optional keep_nyquist bn optional gse or pme specific options The far interactions are computed by using an Ewald mesh calculation The built in methods support both smooth particle mesh Ewald PME and k space Gaussian split Ewald k GSE according to the type parameter In these 60 Chapter 7 Calculating Force and Energy Desmond Users Guide Release 3 4 0 0 7 methods particle charges are spread onto a three dimensional Cartesian mesh and a Poisson equation is solved on this mesh The resulting potentials are used to compute the forces and energy of each particle The Poisson equation is solved efficiently using fast Fourier transforms The splitting parameter o first referenced in Van der Waals and electrostatic interactions determines the far electr
3. 7 2 8 Pair 12 6 terms Pair terms in pair_12_6_es allow for modifying the normally calculated nonbonded interactions either by scaling the interaction energy or by specifying new coefficients to use for a particular pair This partial or modified energy is calculated in addition to the normally calculated interaction energy The functional form of the pair potential is Qij bij Qij p ij rl2 76 Pie ij ij ij The aij bij and qij coefficients are specified in the pair_12_6_es table Table 7 10 Schema for the pair_12_6_es table name type description aij FLOAT scaled LJ12 coeff in ENERGY LENGTH bij FLOAT scaled LJ6 coeff in ENERGY LENGTH qij FLOAT scaled product of charges in CHARGE po INTEGER Ist particle pl INTEGER 2nd particle Pair terms contribute the van der Waals interaction to the pair_vdw Hamiltonian category and the electrostatic interaction to pair_elec 7 2 9 Flat bottomed harmonic well Desmond supports a variant of the usual harmonic stretch angle improper and position restraint terms in which a region of the potential near the equilibrium position is flat All flat bottomed potentials transition to the harmonic region with a continuous first derivative i e forces are everywhere continuous The Hamiltonian categories of the flat bottomed terms are correspondingly st retch_fbhw angle_fbhw improper_fbhw and posre_fbhw The flat bottomed harmonic stretch term su
4. 9 3 RESPA Timestep scheduling is configured as shown in integrator respa near_timesteps in far_timesteps if outer_timesteps i Most Desmond integrator types and force configurations support a splitting of the force field into three computa tional categories with separate scheduling of each The divisions are bonded nonbonded near van der Waals and short range electrostatic interactions and nonbonded far long range electrostatic interactions Additionally certain dynamical events typically corresponding to the dynamics of extended variables such as a thermostat occur outside of a complete NVE step The scheduling of these different categories is controlled by these values During the course of a simulation positions and momenta are updated according to the velocity Verlet algorithm Bilt 64 2 pi t fi t d 2 9 1 rilt 64 Tilt pi t 0 2 0 ms 9 2 Pilt 5 Pilt 5 2 Filt 5t 5 2 The force is split into three components f t f t t t where each of f and f is computed every dt Inde and fd units of time based on the current value of r t fori f phases of time repeated 1 i f times for a total of io phases or an outer time step of A 7 6 It is required that divide 7 and that i divide to Another way to think of this is that unless otherwise specified each full time step is built from a sequence of 7 po sition updates int
5. Table 10 3 Configuration for alchemical name description lambda vdwA values to parameterize the Lennard Jones interactions in the A state List of 0 lt Real lt 1 lambda vdwB values to parameterize the Lennard Jones interactions in the B state List of 0 lt Real lt 1 lambda qA values to scale the partial charges in A state List of 0 lt Real lt 1 lambda qB values to scale the partial charges in B state List of O lt Real lt 1 lambda qc values to scale the partial charges in C state Optional by default there are no C state charges List of 0 lt Real lt 1 lambda bondA values to scale the bond terms in A state List of 0 lt Real lt 1 lambda bondB values to scale the bond terms in B state List of 0 lt Real lt 1 10 1 Configuring free energy simulations 97 Desmond Users Guide Release 3 4 0 0 7 98 Chapter 10 Free Energy Simulations CHAPTER ELEVEN ENHANCED SAMPLING AND UMBRELLA SAMPLING 11 1 Introduction 11 1 1 Who should read this chapter This document is intended to provide all the information needed for a Desmond user to perform umbrella sampling and metadynamics using the enhanced sampling plugin Basic understanding of the theory of umbrella sampling and metadynamics is assumed Though the information in this document will be of interest to developers the primary developer documentation is the Doxygen comments in the sourc
6. delta_max Amax Desmond provides two Brownian dynamics integrators whose primary purpose is to equilibrate systems which might be in high potential energy configurations due to system preparation artifacts brownie_NVT and brownie_NPT They differ in that the latter will sample global cell dimensions as well as particle positions Mathematically the dynamics of these integrators are no different from that of the corresponding Langevin integrators L_NVT and L_NPT of L_NVT Langevin constant volume and temperature and L_NPT Langevin constant pressure and temperature in the limit as T 7 0 In this limit all inertial information is lost and the equations proceed as either V_NVE or Piston_NPH dynamics with particle and piston momenta being sampled independently from Maxwell Boltzmann distributions While it is possible to obtain the mathematical behavior of these integrators by taking T 7 0 in previously discussed integrators obtaining samples from the same stationary distribution the Brownian dynamics integrators have been modified to stabilize the equilibration process from starting points with very large potential energies and forces Specifically all particle and piston velocities are clipped so that no particle is displaced by more than a length of Amax in any direction on position update We typically set Amax 0 1A This additional safety feature prevents run away particles or a collapsing exploding global cell durin
7. input molfile mae read foo mae output molfile pdb write foo pdb atoms input atoms output frame input frames next output close Write every 10th frame in a dtr to a trr input molfile dtr read big dtr output molfile trr write out trr natoms input natoms for i in range 0 input nframes 10 output frame input frame i output close All data is read to and from mol file objects in terms of a small number of classes defined within the module e Atom Represents fixed particle attributes i e no position or velocity Atoms hold references to other atoms through their bonds member use Atom addbond and Atom delbond to change the bond topology e Frame Data from a single timestep Contains position velocity unit cell and physical time e Plugin For each supported file type e g pdb mae trr there is a Plugin object with that name in the module A Plugin can be queried for its capabilities using its can_ methods Nearly all plugins can read files but only some can write Use the Plugin read method to create a Reader and Plugin write to create a Writer Some plugins e g psf read only structure data atoms while others e g dtr read only coordinate data frames If you try to read atoms from a dtr or frames from a psf you ll get an error e Reader A Reader is a handle to an open file Use the atoms member to fetch the atomic structure fro
8. Plugin read path new Reader for given path Plugin write path new Writer for path supply natoms or atoms 118 Chapter 13 Trajectory Format and Analysis CHAPTER FOURTEEN APPENDIX UNITS This appendix explains how numbers provided as configuration parameters are interpreted Many configuration parameters are real numbers that are interpreted as dimensioned quantities Desmond code uses the 2002 CODATA adjustment for units as given by the National Institute for Standards COD 2002 For a given dimension Desmond always uses the same kind of units e Time is in picoseconds ps e Length is in Angstroms A e Energy is in thermochemical kilocalories per mole e Pressure is in Bar e Temperature is in degrees Kelvin K e Charge is in multiples of the absolute electron charge Boolean values are either true or false Integers are assumed to be in the range of two s complement 32bit representations Real valued quantities can be given in decimal or scientific e notation 4 infinity and 4 tinf are also recognized real values as is tnan See st rtod 3 for a full description of the acceptable values for real valued quantities 119 Desmond Users Guide Release 3 4 0 0 7 120 Chapter 14 Appendix Units CHAPTER FIFTEEN APPENDIX CONFIGURATION SYNTAX This appendix describes the configuration file syntax All Desmond applications are configured by mean
9. f wW Configuration files are divided into sections which can in turn contain other sections parameters occur at all levels When discussed in the context of their particular section configuration parameters appear by name in a typewriter font thus plugin When discussed outside of the context of their sections however configuration parameters appear as a keypath in which the name of each enclosing section appears in order from outermost to innermost separated by Desmond Users Guide Release 3 4 0 0 7 periods For example force nonbonded far sigma refers to the sigma configuration parameter in the far subsection of the nonbonded subsection of the force section of the configuration file 2 4 About the equations The equations in this document are concerned with scalars vectors and matrices of various sorts To help clarify the type of a quantity equations in this manual use the following conventions An upper or lowercase letter without bolding or arrows such as A or a is a scalar e An arrow over a variable such as A or G indicates three variables as a three dimensional vector e A boldfaced lowercase letter such as a is a vector of unspecified dimension with a indicating the it element of the vector e A boldfaced uppercase letter such as A is a matrix of unspecified dimensions though usually 3 x 3 with Aj being the element of row and column 7 in matrix Certain quantities that are 3n
10. ci 7 j 2 Note the astute reader may observe that our equations vary from the original MTK equations in the handling of the 1 N terms in the p Di and 7 equations Recalling the definitions of the invariant scalar and phase space density from Piston_NPH see Piston_NPH constant pressure and enthalpy the above ODE preserves the scalar H x s C p 1 4 p t 8 p n Dates 2w ae and the phase space density Q exp Y BT Y CIG aari J i ij In particular if T Tk Tp T then the density shown below is preserved Q exp kgT S P 1 dm 2w a ai 9 6 Dynamical systems 81 Desmond Users Guide Release 3 4 0 0 7 Table 9 9 Configuration for MTK_NPT description Used to set the mass see Piston_NPH Time gt 0 Equilibrium temperature see Piston_NPH and NH_NVT Optional defaults to the global reference temperature Temperature gt 0 Description of the barostat chain see NH_NVT Nos Hoover chain Description of the particle thermostat see NH_NVT Nos Hoover chain name barostat tau barostat T_ref barostat thermostat thermostat 9 6 7 L_NPT Langevin constant pressure and temperature The L_NPT dynamical system is configured as shown in integrator L_NPT barostat tau Tp T_ref T thermostat tau Tb seed S thermostat tau T seed s The Langevin constant pressure and
11. fepio_pairmaps 36 ct ffio_ct_type lines omitted solvent m atom 915 s_m_pdb_atom_name s_m_pdb_residue_name s_m_chain_name i_m_residue_number r_m_x_ coord r_m_y_coord r_m_z_coord i_m_atomic_number OWS SOL HWS SOL HWS SOL OWS SOL HWS SOL x K K K WWNN DY Dd 004 WN EH se lines omitted lines omitted 5 690000 12 750000 11 650000 4 760000 12 680000 11 280001 5 800000 13 639999 12 090000 15 549999 15 110001 7 030000 14 980000 14 950000 7 840000 HWS SOL X 3 14 960001 15 210000 6 230000 and will vanish Fr ORF 142 Chapter 18 Legacy Applications Preparing a Maestro structure file CHAPTER NINETEEN ENHANCED SAMPLING FUNCTION REFERENCE Class Binary Threaded Arguments e a array e b array Return a x b computed element wise by the binary threading rules Class Binary Threaded Arguments a array e b array Return a b computed element wise by the binary threading rules Class Binary Threaded Arguments a array e b array Return a b computed element wise by the binary threading rules Class Binary Threaded Arguments a array e b array Return a b computed element wise by the binary threading rules A 143 Desmond Users Guide Release 3 4 0 0 7 Class Binary threaded Arguments a array e b integer Return a performed element wise by the binary threading rules Note that b will be rounded to get an integer
12. met s_fepio_name i_fepio_stage thane_to_ethan 1 m_atom 8 lines omitted 140 Chapter 18 Legacy Applications Preparing a Maestro structure file Desmond Users Guide Release 3 4 0 0 7 feror ftE lines omitted lines omitted fo met 4 s_fepio_name i_fepio_stage thane_to_ethan 2 m_atom 8 lines omitted ffio_ff lines omitted fepio_fep s_fepio_name i_fepio_stage thane_to_ethan a fepio_atommaps 13 i_fepio_ai i_fepio_aj 111 222 333 4 4 1 The 4 5 6 7 and 8th atoms in state A vanish and become dummy atoms in state B panet 6 1 8 1 9 4 The 4 5 6 7 and 8th atoms in state B don t exist and are dummy atoms in state A 10 10 5 11 106 12 12 7 13 13 8 wo OAD UO fepio_bondmaps 12 i_fepio_ti i_fepio_tj i_fepio_ai i_fepio_aj 11112 22213 3 3 1 1 4 The bond between atoms 1 and 4 in state A does not exist in state B but will not be changed lines omitted fepio_anglemaps 23 18 2 Preparing the structure file for Free Energy Simulations 141 Desmond Users Guide Release 3 4 0 0 7 Ey s_ i_fepio_ti i_fepio_tj i_fepio_ai i_fepio_aj i_fepio_ak 1 1 1 6 5 1 The angle between atoms 6 5 1 in state A does not exist in state B but will not be changed 22 075 1 The angle between atoms 7 5 1 in state A does not exist in state B lines omitted fepio_dihedmaps 18 fepio_exclmaps 78
13. p The energy is broken down into components and printed in columns of the eneseq file indexed by simulation time column t ime conserved The sum of potential kinetic and extended system energy E Ej Ep Ez usually it is possible that there are additional internal energy terms added to the conserved quantity but currently this only arises in the case of some polarization schemes For many integration methods this quantity is asymptotically conserved as the simulation timestep goes to O and serves as a check on the correctness of the trajectory column E potential The value of U r column e E p kinetic The value of K p gt Ip 1 2m column E k extended The energy associated with the extended variables of the dynamical system being integrated column E_X center of mass The value of gt gt Pill 2 Xo Mmi which is the center of mass kinetic energy of the entire system column E_c To the extent that the system forces violate Newton s third law and no steps are taken to 4 6 Configuring the built in plugins 31 Desmond Users Guide Release 3 4 0 0 7 periodically remove center of mass motion this quantity can grow over time e force correction The value of 67 gt fill2 8m where f is the force on particle i Because its addition to the energy gives an exactly up to arithmetic conserved quantity in systems where the potent
14. 0 7 Table 4 10 Configuration for BiasingForce name description cm moi Biasing force definition for a list of cm_moi groups List groups The groups to restrain List displace_coeff Force coefficients for displacement restraints List of 3 Energy Length displacement Relative displacement between the two groups List of 3 Lengths distance_coeff Force coefficient for distance restraint Energy Length distance Distance between the two groups Length orient_coeff Force coefficients for orientational restraints List of 3 Energies Euler_angles Euler angles between the two groups List of 3 Degrees use_lab_frame_for_displacement If true the displacement between the groups are measured in the reference frame of the simulation box Boolean pull_displacement Velocity of pulling in displacement List of 3 Length Time pull_distance Velocity of pulling in distance Length Time pull_Euler Velocity of pulling in orientation List of 3 Degree Time to The time to begin apply pulling as specified by pull_displacement pull_distance and pull_Euler This is not to be confused with the biasing force itself which is applied from the beginning of the simulation Time output first The time to write the first biasing results Time output interval The interval at which to write the biasing results Time output name The name of the file to
15. 04 5 13 3 4 fskeycheck Occasionally frameset files can be corrupted on disks The fskeycheck tool will check the integrity of the timekeys and frame files Using the fix option will output a new timekeys file in the current working directory that truncates any bad frames The frameset can be updated by replacing the original timekeys file with the newly generated one fskeycheck fix framesetdir 13 3 5 rebuild_timekeys The information in the timekeys file is redundant It is used to make a quick association between times and the bytes that represent the associated frames in the frame files If the timekeys file is corrupt broken or missing the rebuild_timekeys tool will scan all the frame files and create a new timekeys file in the current working directory 13 3 Command line tools for framesets 113 Desmond Users Guide Release 3 4 0 0 7 rebuild_timekeys framesetdir 13 4 Python tools for trajectories and framesets Command line tools are useful for a quick look at the data contained in trajectories but it is difficult to write analysis tools from the text tools or the raw format itself Desmond provides a library of C and Python tools to access frame data The Python modules make it easy to write high performance scripts to analyze trajectory data The data are accessible via numpy arrays 13 4 1 framesettools module for direct access The framesettools module allows Python scripts access to the raw
16. 1 1 5 41 0 403379 0 990407 0 655399 H5 1 1 6 42 0 858372 0 916697 0 724639 HH 1 1 lines omitted f m_ct ae lines omitted s_ffio_ct_type a lines omitted solvent m_atom 2484 i_m_mmod_type r_m_x_ coord r_m_y_coord r_m_z_coord s_m_pdb_atom_name i_m_atomic_number i_ffio_grp_ligand 1 16 7 429789 7 792630 4 945186 OW 6 0 2 42 6 709420 8 366949 4 498097 Hwi 1 0 3 42 7 200478 6 819860 4 736009 Hw2 1 0 lines omitted 18 2 2 Alchemical free energy simulations The structure file used for alchemical free energy simulations consists of the following components e original_ct system in state A e perturbed_ct system in state B e environment_ct s The original_ct contains the unperturbed version of the molecule in the alchemical transformation and the per turbed_ct contains what the original_ct becomes They both contain ffio information to describe the force field parameterization in their respective states They also both contain FEPIO information specific to alchemical free energy simulation The environment_ct s are component CTs or multicomponent CTs that do not undergo alchemical transformation These CTs have only f fio information but not FEPIO information CT level MMFEPIO properties Both the original_ct and the perturbed_ct must contain a user specified name for the FEP transformation and whether it corresponds to the original or the perturbed state 18 2 Preparing the str
17. 133 136 143 155 20 2 Licensed Companion Softwares soca os aaa aa o e ct i Ss 158 Bibliography 163 Desmond Users Guide Release 3 4 0 0 7 Release 0 6 0 Date December 18 2012 Contents CONTENTS 1 Desmond Users Guide Release 3 4 0 0 7 2 CONTENTS CHAPTER ONE FRONT MATTER 1 1 Notice The Desmond User s Guide and the information it contains is offered solely for educational purposes as a service to users It is subject to change without notice as is the software it describes D E Shaw Research assumes no respon sibility or liability regarding the correctness or completeness of the information provided herein nor for damages or loss suffered as a result of actions taken in accordance with said information No part of this guide may be reproduced displayed transmitted or otherwise copied in any form without written authorization from D E Shaw Research The software described in this guide is copyrighted and licensed by D E Shaw Research under separate agreement This software may be used only according to the terms and conditions of such agreement 1 2 Copyright 2012 by D E Shaw Research All rights reserved 1 3 Trademarks Ethernet is a trademark of Xerox Corporation InfiniBand is a registered trademark of systemI O Inc Intel and Pentium are trademarks of Intel Corporation in the U S and other countries Linux is the registered trademark of Linus Torvalds in the U S
18. Pastor and B R Brooks Constant pressure molecular dynamics simulation The Langevin piston method J Chem Phys 103 4613 4621 September 1995 Gun 1984 W F van Gunsteren H J C Berendsen Molecular dynamics simulations Techniques and approaches In A J Barnes et al editor Molecular Liquids Dynamics and Interactions NATO ASI C 135 pages 475 500 Reidel Dordrecht The Netherlands 1984 Mar 1992 G J Martyna M L Klein and M Tuckerman Nose Hoover chains The canonical ensemble via continuous dynamics J Chem Phys 97 2635 2643 August 1992 Mar 1994 G J Martyna D J Tobias and M L Klein Constant pressure molecular dynamics algorithms J Chem Phys 101 4177 4189 September 1994 COD 2002 P J Mohr and B N Taylor CODATA recommended values of the fundamental physical constants 2002 Rev Mod Phys 77 1 1 107 2005 Rei 1994 Sebastian Reich Momentum conserving symplectic integrators Physica D 76 373 383 1994 Sha 2005 Yibing Shan John L Klepeis Michael P Eastwood Ron O Dror and David E Shaw Gaussian split Ewald A fast Ewald mesh for molecular simulation J Comp Phys 122 5 2005 Shi 2003 Michael R Shirts Eric Bair Giles Hooker and Vijay S Pande Equilibrium free energies from nonequi librium measurements using maximum liklihood methods Physical Review Letters 91 2003 sqlite http www sqlite org SQLite home pa
19. a multigrator configuration that corresponds to one of the previous integrators NH_NVT MTK_NPT L_NPT etc approaches the results of that integrator with barostat r parameters multiplied by ng and thermostat 7 parameters multiplied by nr The Langevin thermostat steps evolve the p variables according to 5 Bi 9 015 t ng7 where each component of the random vector S t is a standard Wiener process The NoseHoover thermostat steps add Nos Hoover chains consisting of 2 extended variables 7 vi for each rI governing the particles of thermostat j and evolve them and the p variables according to G u jwi Bi BO Jo pill mi Ci vivu x i 4 wl c vivi a Us a Df VA Ww091 Ch where wl c npr with cl kgT Nj and Cs kpT where Nj is the number of degrees of freedom of the governed particles 7 The Antithetic thermostat steps apply the discrete antithetic thermostatting dynamics described in Anti_NVT Antithetic constant volume and temperature The Mixed thermostat applies two thermostatting dynamics symmetrically Specifically the thermostat step at the beginning of the sequence of NVE steps applies the two thermostats consecutively in the order they are listed in the Mixed type field The thermostat step at the end of the NVE steps applies them in the reverse order The consituent thermostats are configured exactly as they otherwise would be albeit placed within the M
20. and other countries All other trademarks are the property of their respective owners Desmond Users Guide Release 3 4 0 0 7 4 Chapter 1 Front matter CHAPTER TWO PREFACE 2 1 Intended audience This guide is intended for computational scientists using Desmond to prepare configuration and structure files for molecular dynamics simulations It assumes a broad familiarity with the concepts and techniques of molecular dy namics simulation 2 2 Prerequisites Desmond runs on Intel based Linux systems with Pentium 4 or more recent processors running CentOS 5 4 RHEL5 or later Linux clusters can be networked with either Ethernet or InfiniBand To build the source code Desmond is known to work with gcc Version 4 5 2 and glibc Version 2 5 Certain python scripts require a recent version of Python 2 version 3 is not supported we recommend Version 2 7 1 or greater This guide assumes that someone has prepared the Desmond executable for you either by installing a binary release or by building the executable 2 3 Format conventions Command lines appear in a typewriter font in some cases bolding draws your attention to a particular part of the command desmond includ quil cfg Placeholders intended to be replaced by actual values are obliqued desmond tpp 4 restore checkpoint_file Configuration file examples also appear in a typewriter font mdsim title last_time t checkpt 1 plugin
21. b and c are covered by an AH2 constraint in the A state and particles A d and e are covered by an AH2 constraint in the B state then the set of constraint terms in the alchemical DMS file should include an AH4 constraint between A and b c d and e 17 3 4 Virtual sites No support is offered for alchemical virtual sites at this time 17 3 5 Polar terms No support is offered for alchemical polar terms at this time 132 Chapter 17 Appendix DMS file format CHAPTER EIGHTEEN LEGACY APPLICATIONS PREPARING A MAESTRO STRUCTURE FILE As discussed in Input Desmond requires two files for input a structure file that defines the chemical system and a configuration file that sets simulation parameters The details of setting configuration parameters are described in Running Desmond This chapter describes how Desmond prior to version 2 4 specified the structure file 18 1 Format A structure file also known as a Maestro file or MAE file file name suffix mae is organized as a set of nested blocks Each block has a set of attributes and can contain other blocks Some blocks called arrays or indexed blocks contain multiple records Blocks start and end with curly braces Within each block attribute names are listed first followed by and finally the values of those attributes A typical structure file starts with an unnamed block as shown in s_m_m2io_version 2 0 0 The unnamed block specifies the vers
22. both are true To control the complexity of the output users can pick a maximum depth of the tree view dmax and a minimum number of occurrences Cmin below which the activity is not reported For example most initialization activities occur just once SO Cmin 2 Suppresses them When profiling a simulation run in parallel profile prints the profile for process 0 If the simulation is sufficiently load balanced this is representative of the whole computation Table 4 22 Configuration for profile name description max_depth Maximum depth of the tree view Optional by default infinite Integer min_calls Minimum number of occurrences to report Optional by default 1 Integer show_tree Whether to display the tree view Optional by default true Boolean show_flat Whether to display the flat view Optional by default true Boolean 38 Chapter 4 Running Desmond CHAPTER FIVE THE GLOBAL CELL This chapter discusses Desmond s parallelization strategies and describes how to configure the global cell 5 1 Parallelization As described in Space the global cell is Desmond s representation of the space occupied by the chemical system It fills an infinite volume by tiling the space periodically with repetitions of the global cell To parallelize the computations Desmond decomposes the work spatially Therefore configuring the global cell appropriately requires an understanding o
23. by default empty List of names nonbonded configuration of the nonbonded forces Can be set to none for no nonbonded forces configuration term configuration of a set of special force terms provided typically by a general force plugin configuration ignore_com_dofs A user assertion that at least up to exact arithmetic the dynamics do not have any net center of mass motion Boolean 50 Chapter 7 Calculating Force and Energy Desmond Users Guide Release 3 4 0 0 7 7 1 1 Force terms This section is a specification of force term plugins similar in layout to the application specific plugin section force term list ees Key sesx J key type type term specific configuration options Examples of such terms that we have seen so far are BiasingForce described in Biasing Force and e_bias described in e_bias 7 2 Bonded pair and excluded interactions This section describes the built in bonded term objects that can be used in a Desmond application specified by records in the bond_term table of the DMS file 7 2 1 Stretch terms The vibrational motion between two atoms i 7 is represented by a harmonic potential as Velraj felrig ro where fe is the bond force constant in units of Energy Length and ro is the equilibrium bond distance Terms in stretch_harm are evaluated using this potential Table 7 2 Schema for the stretch_harm table name type
24. checkpt first tf interval ti name p write_first_step by write_last_step b 20 Chapter 4 Running Desmond Desmond Users Guide Release 3 4 0 0 7 Setting checkpt none shuts off checkpointing A checkpoint is written at simulation time and thereafter with a period t or at the wall clock interval tu as measured from the start of each invocation of the simulator The output file name convention is followed for the checkpoint files see Naming output files You can cause mds im to write a checkpoint file initially and finally by setting b and bf respectively to true Table 4 3 Configuration for checkpointing name description first First time to create a checkpoint time interval Periodic interval at which to create checkpoints time wall_interval Periodic interval at which to create checkpoints wall clock time in units of seconds time name Output filename to use for the checkpoint files filename write_first_step Whether to write a checkpoint file before the first step is taken Boolean write_last_step Whether to write a checkpoint file after the last step is taken Boolean 4 4 2 remd The remd application in Desmond implements the replica exchange protocol sometimes known as parallel tempering The number of replicas that can be simulated is limited only by the number of processors available and that an equal number of processors must be assigned to
25. cutoff If the collective variables in the current configuration are more than cutoff number of kernel widths away from the center of a kernel the kernel is not computed If the cutoff is 0 0 an infinite cutoff is used first determines the first time at which a Gaussian is added interval determines the time between Gaussian drops A value of 0 0 indicates that a Gaussian is dropped on every time step name If non empty this gives the name of the kernel sequence file which logs every kernel added to the simulation See below initial If non empty gives the location of a file containing kernels to be added at the beginning of a simulation See below All kernels that are added to the simulation are logged to the kernel sequence file where each kernel is described by the time it was added its height and its widths Lines that begin with a hash linecode are comments This same format may be used to define an initial kernels file which is loaded at Desmond boot The logged kernels can be used to initialize a new simulation with the metadynamics potential produced by a previous simulation or to start the simulation with a potential defined by an arbitrary kernel mixing model When the kernels are loaded the time values are required but are ignored in the computation all initial kernels are used regardless of the current value of chemical time All initial kernels are written to the kernel sequence file before any new kernels are wr
26. families of related functions or blocks of code dedicated to a particular task 4 7 Configuring optional sections 37 Desmond Users Guide Release 3 4 0 0 7 This feature is primarily to assist developers the activities are described by short names that are somewhat self explanatory The profile section is optional If omitted no profile is generated Profile configuration is given in profile show_tree b show_flat by min_calls Cmin max_dept h dmax Two profile views can be output a tree view and a flat view both analogous to the output of the well known profiler gprof The tree view gives times for various activities in a hierarchy since activities can contain sub activities or children The accumulated time for each activity is the total time spent in that activity and its children An activity can occur in more than one place in the hierarchy The flat view removes the hierarchy and lists one line per activity accumulating times spent in an activity which may be the result of different parent activities Additionally the time printed for the flat view is given as the difference between the time spent in that activity and the total time spent in the children of that activity and hence the total time in the flat view should be roughly equal to the total runtime of the application minus some startup and shutdown overhead The Boolean variables b and bf control which views are presented By default
27. n The desired free energy difference between Ho and H is then given by AF AFo 1 n ZE AFin 1 n 1 Table 10 1 Configuration for FEP name description type The type of free energy simulation to run alchemical binding alpha_vdw The parameter in the softcore potential Real gt 0 window Selecting the values to use in this simulation Integer 0 1 n output first The time to write the first energy difference value Time gt 0 output interval The interval at which to write the energy difference estimates Time gt 0 output name The name of the file to which to write the energy estimates Filename 10 1 1 Bennett acceptance ratio method Consider a simulation under Hamiltonian A and another under Hj both at temperature T N samples of wo H r H r are accumulated in the former simulation and M of W H r H r in the latter The free 94 Chapter 10 Free Energy Simulations Desmond Users Guide Release 3 4 0 0 7 energy difference between systems a and b is estimated by solving the following nonlinear equation for AF Na Ne 3 y l s i 1 Reexp G Wi AF Fz 1 Re exp G AF WI b Charles Bennett see Ben 1976 first demonstrated that this solution provides the minimum variance estimate of AF Two decades later Michael Shirts et al see Shi 2003 proved that it is also the maximum likelihood estimator of AF The Bennett acceptan
28. reference_tim Start time for the simulation Time r_clone Radius of particle home box visibility Length gt 0 margin A user promised maximum displacement of any particle between migration events Length gt 0 partition Number of process subdivisions along each axis Optional by default 0 0 0 meaning that Desmond sets them List of three Integers est_pdens Average number of particles per unit volume Optional by default computed from the structure file 1 Length gt 0 5 3 Migration Migration is configured as shown in migration first tf interval ti Desmond partitions particles across processors by a spatial decomposition As such when particle positions change home box ownership must be recalculated and interprocess communication must occur to make each process aware of new particles in its view This is called migration Since it is a significant computational and communicative task which need not be performed at every position update this task is scheduled independently of position changes The parameters ts and t set the time for the first migration update and the interval of time between later migration update 5 3 Migration 43 Desmond Users Guide Release 3 4 0 0 7 Table 5 2 Configuration for migrate name description first Approximate amount of time of the first migration Time gt 0 interval Approximate amount of time between subsequen
29. the electrostatic interactions are split into those between particles within the cutoff radius and those between more distant particles Modified electrostatic interactions are computed explicitly for the closer particle pairs while the distant particle pairs are computed according to a more efficient method thus further improving performance Interactions between pairs of particles separated by less than the cutoff radius are called nonbonded near interactions or more briefly the near interactions They comprise both van der Waals forces and the short range electrostatic forces 8 Chapter 3 Key Concepts Desmond Users Guide Release 3 4 0 0 7 Electrostatic forces between pairs of particles separated by more than the cutoff radius are referred to as nonbonded far interactions or far interactions Instead of computing each pair wise interaction explicitly Desmond computes far interactions more efficiently in Fourier space thus 1 The application maps charges from particles to nearby grid points needed for the Fourier transform charge spreading 2 Using this charge density it determines the nonbonded far potential at each mesh point via Fourier space tech niques 3 It calculates the resulting forces on the particles from the results at the nearby grid points force interpolation Even with optimizations such as the Fourier space computation far interactions are expensive to compute Because the overall force these interacti
30. type none Verlet PLS thermostat type Langevin NoseHoover Antithetic Mixed timesteps nr Langevin Lau 7 seed Ss NoseHoover mts m tau 71 Tn Antithetic bath_dof Np seed SA Mixed type Langevin NoseHoover Antithetic Langevin NoseHoover Antithetic barostat type MTK timesteps NB MTK T_ref T tau Tp thermostat type none Langevin NoseHoover Antithetic Langevin tau Th seed Sp NoseHoover mts m _ b b tau Ti Ta Antithetic bath_dof NP seed so The multigrator is an integrator developed in house to allow greater flexibility in the design of the integration step combining the features of the dynamical system and stochastic integrators It also allows the user the option of carrying out thermostat and barostat updates less frequently than once per outer RESPA timestep reducing the performance overhead of extended system dynamics The integration update steps vary periodically with a full period of ng updates spanning a chemical time of ng 9 6 Dynamical systems 87 Desmond Users Guide Release 3 4 0 0 7 Every update contains a full NVE step which updates positions and momentum to approximate the solution of P p m Pi Vr U r according to the selected nve t ype and the integrator respa schedule only certain RESPA schedules are currently compati
31. 0 62 1 00 Verlet 3 3 3 1 3 0 35 1 00 Verlet 1 4 4 1 7 0 46 1 00 PLS 1 1 1 3 7 1 00 1 00 PLS 1 1 2 2 3 0 62 0 62 PLS 1 1 3 2 7 0 73 0 73 PLS 1 1 4 25 0 68 0 68 PLS 1 2 2 2 3 0 62 0 74 PLS 1 3 3 2 3 0 62 1 00 PLS 1 3 6 1 5 0 41 0 65 PLS 1 2 4 2 1 0 57 0 68 PLS 1 4 4 1 7 0 46 1 00 Empirical estimation of the maximum before the onset of instability for various types of NVE step The last two columns give the fraction of each method s relative to that of Verlet 1 1 1 and to Verlet with the same schedule 90 Chapter 9 Dynamics Desmond Users Guide Release 3 4 0 0 7 9 6 12 The Concatenator integrator integrator Concatenator sequence name key1 type type1 time T os A name keyn type typen time T y keyi keyn 1 The concatenator is a means to alternate different integrator types for various periods of time in a cyclic sequence The sequence parameter specifies the integrator types type to be run for periods of time T The sequence is treated cyclically starting over from the first one after the last one finishes The configuration section for each integrator is given by an arbitrary key name key in the remainder of the concatenator configuration For example if one wished to alternately employ L_NVT and V_NVE for periods of 100 ps and 500 ps each one configures this integrator as follows integrator Concatenator sequence name fir
32. 2 Retone Note Ordinarily near interactions restrict the size of the clone radius more than any other consideration For all restrictions on the size of the clone radius see Appendix Clone Radius Restrictions For setting these three radii the 5 1 Parallelization 41 Desmond Users Guide Release 3 4 0 0 7 following rule of thumb is useful for most simulations 1 Choose the cutoff radius Reut 2 Choose the margin A 3 Set the clone radius Relone to half of Riazy Reut A 5 2 Configuration Configuring the global cell involves setting e the reference time and e the clone radius In addition if you re running Desmond in parallel you can also e specify how you wish to partition the global cell among the processes and e provide an estimate of average particle density per homebox These parameters are discussed below The global cell s section in the configuration file appears as shown in the following Synopsis Configuring the global cell global_cell reference_time tr r_clone Rolone partition ni na ns margin A est_pdens d The global cell is centered at the origin with edge vectors given by the lattice vectors G b Z read from the structure file This is described Global cell The global cell is responsible for the time coordinate t initialized to t The integers n1 ng ng specify how the global cell is partitioned among processes with each process assigned a
33. Global cell component Attribute X component of a vector r_chorus_box_ax Y component of a vector r_chorus_box_ay Z component of a vector r_chorus_box_az X component of b vector r_chorus_box_bx Y component of b vector r_chorus_box_by Z component of b vector r_chorus_box_bz X component of c vector r_chorus_box_cx Y component of c vector r_chorus_box_cy Z component of c vector r_chorus_box_cz 134 Chapter 18 Legacy Applications Preparing a Maestro structure file Desmond Users Guide Release 3 4 0 0 7 Note Each ct block in a structure file must contain the same global cell specification as every other ct block in that file if any 18 1 2 Particles and pseudoparticles After loading the structure file Desmond scans the ct blocks looking for particles to include in the simulation Each ct block must contain one or more atoms depending on the force field to be used it can also contain pseudoparticles representing additional charge sites Pseudoparticles are described in general in Force fields specific implementa tions are described in Virtual sites The atoms in a ct block are specified by the records in the m_atom array Pseudoparticles if any are given by the records in the fio_ pseudo array within the io_ ff subblock of the ct block Each atom and pseudoparticle record can contain any number of attributes however Desmond reads only e the positions an
34. If this is not the desired behavior then pow should be used instead acos Class Threaded Arguments a array Return the element wise arccosine of a note that this function is not numerically stable for arguments near 1 or 1 angle Class Normal Arguments e a 3 element array e b 3 element array Return The cosine of the angle between a and b This function does not return the angle directly due to numerical issues that arise due to the periodicity of angles In particular inverse trigonometric functions often have singularities in their derivatives angle_gid Class Normal Arguments e pl particle e p2 particle e p3 particle Return The cosine of the angle of p1 p2 and p3 This function does not return the angle directly due to numerical issues that arise due to the periodicity of angles In particular inverse trigonometric functions often have singularities in their derivatives angle_gid_radians Class Normal Arguments e pl particle e p2 particle e p3 particle 144 Chapter 19 Enhanced sampling function reference Desmond Users Guide Release 3 4 0 0 7 Return Angle of a and b in radians Result is in the range 0 7 This function is not safe to use if the angle is near 0 or 7 because the derivative of this function diverges It is preferable to use the function angle when possible angle_radians Class Normal Arguments e a 3 element array e b 3 element array Return Angle of a and
35. NON INFRINGEMENT IN NO EVENT SHALL THE COPYRIGHT HOLDERS OR ANYONE DISTRIBUTING THE SOFTWARE BE LIABLE E 158 Chapter 20 Licenses and Third Party Software Desmond Users Guide Release 3 4 0 0 7 FOR ANY DAMAGES OR OTHER LIABILITY WHETHER IN CONTRACT TORT OR OTHERWISE ARISING FROM OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE T 20 2 2 SunPro Error Function Desmond uses an implementation of the error function that includes the following notice Copyright C 1993 by Sun Microsystems Inc All rights reserved Developed at SunPro a Sun Microsystems Inc business Permission to use copy modify and distribute this software is freely granted provided that this notice is preserved 20 2 3 ANTLR LICENSE Copyright c 2005 2008 Terence Parr All rights reserved Redistribution and use in source and binary forms with or without modification are permitted provided that the following conditions are met 1 Redistributions of source code must retain the above copyright notice this list of conditions and the following disclaimer 2 Redistributions in binary form must reproduce the above copyright notice this list of conditions and the following disclaimer in the documentatio
36. S465 4 2h dd EE 11 2 Using the Enhanced Sampling Plugin ss 6 coe eey Re ba ee EE Ee Ee eS NES interpreter cat wena te hw A A Ge eee GB AS oe eee REELS S 11 4 Metadynamics s 5 des deed Ge pada bebe ee ba eee e pha ewe gee eo 11 5 Exampless 0 Se bh be eR OE ee eA eee eee eed eee eee ea an Extending Desmond 121 Implementanom 354 2 ak 2 4 werk ao ee HR SE ES EIA BS Re ae YA ee 12 2 Running your plugins es Saigo gg a ke ne 4S did pd bE ee 4 A Trajectory Format and Analysis 13 1 Structure of frameset directories o e na RAR RR WA e ew Dee A 13 2 SO CatenalOMOPLON s ac Be GO a a ES Bae ae SEES Res be ee A 13 3 Command line tools for framesets ee 13 4 Python tools for trajectories and framesets e Appendix Units Appendix Configuration syntax 15 1 Examples 0 0020 eb Be rl whee a a eii edo be a ba da eee Appendix Clone Radius Restrictions Appendix DMS file format TAI Molecules s opina e a RA A ee FO ee ee A A Legacy Applications Preparing a Maestro structure file 18 1 A haunt AE ae a a Re ew a ee 18 2 Preparing the structure file for Free Energy Simulations Enhanced sampling function reference Licenses and Third Party Software 20 1 Licensing Desmond for Non Commercial Research o o ossec o oo rarere iana 65 67 69 69 70 71 72 73 73 93 93 109 109 110 111 111 112 112 114 119 121 122 125 127 127 129 131 133
37. Software Desmond and its related software makes use of several software packages prepared by organizations and individuals outside of D E Shaw Research We include here the licensing terms for two of those packages 20 2 1 Boost C Libraries Desmond 3 x uses Boost version 1 45 0 available from the Boost website http www boost org under the terms of the Boost Software License Version 1 0 Boost Software Licens Version 1 0 August 17th 2003 Permission is hereby granted free of charge to any person or organization obtaining a copy of the software and accompanying documentation covered by this license the Software to use reproduce display distribute execute and transmit the Software and to prepare derivative works of the Software and to permit third parties to whom the Software is furnished to do so all subject to the following The copyright notices in the Software and this entire statement including the above license grant this restriction and the following disclaimer must be included in all copies of the Software in whole or in part and all derivative works of the Software unless such copies or derivative works are solely in the form of machin xecutable object code generated by a source language processor THE SOFTWARE IS PROVIDED AS IS WITHOUT WARRANTY OF ANY KIND EXPRESS OR IMPLIED INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY FITNESS FOR A PARTICULAR PURPOSE TITLE AND
38. They also change more slowly in time than the other forces For many simulations then you can improve performance by configuring Desmond to compute the far interactions less often for example on alternating timesteps The integrator still computes the near interactions every timestep but it skips the far range computations half the time weighting the results accordingly to compensate for not including them at every timestep Typically near interactions vary at a rate intermediate between bonded forces and far interactions Given their often dominant computational expense Desmond also allows these to be scheduled less often Desmond allows timestep scheduling as follows e Bonded forces are computed at every timestep This is then called the inner timestep e Nonbonded near forces can be computed at every nth inner timestep as configured This is then called the near timestep e Nonbonded far forces can be computed at the same interval as nonbonded near forces or a multiple of it This is then called the outer timestep Timestep scheduling appears as a configuration parameter called RESPA an acronym that stands for reference system propagator algorithm Constraints among particles let you lengthen the timestep by not modeling the very fastest vibrations the integrator moves these constrained particles in unison A variety of geometries can be constrained this way e a fan of 1 8 particles each bonded to a central particle such as
39. To ensure that a given process can access all the data it needs to compute such interactions Desmond copies data for any particle that s outside the home box but within a given distance of its edge Such copies are called clones and this distance is the clone radius For example if particle A near the edge of its home box A participates in a bond with particle B just outside home box A then process A has access to data associated with both particles A which it owns and B which it clones Because each face of the global cell wraps to its opposite particles are also cloned when they are close enough to particles on the opposite face of the global cell as well as the edges of their home box If you re running Desmond serially a single process the home box equals the global cell and this is the only manifestation of clones in the simulation 39 Desmond Users Guide Release 3 4 0 0 7 oa r y SI E gt I i G I I I L I I i j I i I j i I i j i i L L t y i 1 1 LM E G at Meee ewe ee we we we ew ee ew ew ee Figure 5 1 An 2D illustration with nine particles labeled A through I in a 2x1 global cell partitioned between two processes into two homeboxes Below per process views of space with copies of remote and local particles in each processes clone buffe
40. Ues 7 1 These are the bonded van der Waals and electrostatic terms respectively The bonded term arises from the covalent bond structure of the molecules This term includes stretch terms involving two particles connected by a bond angle terms involving three particles connected by two bonds and dihedral torsion terms involving four particles connected by a chain of three bonds During startup Desmond scans the records in the bond_term table and creates a corresponding bonded force term The names of these terms are printed to the log by Desmond during startup The bonded include and 49 Desmond Users Guide Release 3 4 0 0 7 bonded exclude configurations allow you to override the set of bonded terms that will be created These en tries are lists and are empty by default A value of in bonded exclude will turn off all bonded force terms Putting the name of a specific term in bonded exclude will turn off just that term Putting the name of the term in bonded include will override bonded exclude and ensure that the force term gets evaluated Similarly for the include and exclude lists in virtual records from the virtual_term table and constraint records from the const raint_term table as well as polar records from the polar_term table The van der Waals and electrostatic terms are known as nonbonded terms because they include all pairs of particles in the sys tem that are not bonded More precisely they include
41. all pair interactions that are not explicitly excluded by the force field Many force fields also define a scaling for the 14 that is atoms separated by three bonds van der Waals and electrostatic interactions called partial 14 interactions This is a scaling to reduce the strength of these interactions since they are correlated with the bonded terms In Desmond these 14 interactions are implemented in the same way as bonded interactions and it is simplest to think of them in this way For example their interactions are not subject to a distance cutoff and they are treated as bonded terms in multiple timestepping integration Equation 7 1 can now be refined U Upondeat X Uvawt Y quajertelri W20 r j EN 2 5 EN gt qiqjerf ri V20 rij y qiqjerf rij V20 rij 1 5 i j EN where N is the set of pairs that are non excluded pairs The term Upondea includes the partial 14 interactions The second term is the van der Waals term and the remaining three terms comprise the electrostatic term The near nonbonded terms second term and third term for pairs in N are calculated together in the same cutoff limited compute kernel in Desmond and is called the nonbonded near calculation The far nonbonded term fourth term is computed by means of the PME or k GSE algorithms Finally the term fifth term represents the far exclusion which subtracts the far term contribution of excluded pairs Most molecular dynamics force fie
42. and the widths of the gaussian may be an arbitrary expression and the height and widths expression is only evaluated when a gaussian is added to the potential On each gaussian addition the height width and center of the resulting gaussian is written to a file as specified in the declare_meta statement in the header of the potential min_image Class Normal Arguments e a 3 element array Return the minimum image of a with respect to the unit cell mod Class Binary Threaded Arguments e a array 149 Desmond Users Guide Release 3 4 0 0 7 e b array Return mod a b computed element wise by the binary threading rules Answer is between 0 and b including 0 and excluding b ncoordination Class Normal Arguments e r0 scaling distance e n upper exponent e m lower exponent e gids1 first array of gids e gids2 second array of gids Return 5 gt EcrstrrY Where rij is the minimum image distance between particle from the first list and particle j i j 1 rij r0 from the second list returns the coordination number between two groups of atoms norm Class Normal Arguments a array Return Magnitude norm of a Equivalent to sqrt norm2 a norm2 Class Normal Arguments a array Return the dot product of a with itself note that if a is a scalar this is simply the square of the scalar pos Class Normal Arguments e gid particle Return the position of the particle whose G
43. and then sampled on the mesh The Fourier transform then implements a spectral convolution with a kernel Finally forces and energies are accumulated using another B spline convolution in real 7 4 Nonbonded far interactions 61 Desmond Users Guide Release 3 4 0 0 7 space The spectral convolution kernel is that of a Gaussian charge density of width deconvolved twice by the B spline functions It is necessary to choose an order for the B splines 2 for each dimension Orders of 4 7 are supported As a balance between accuracy and efficiency order 5 quartic interpolation is recommended for most applications Table 7 20 Configuration for pme name description order Order of interpolation along each axis List of Integers 4 lt integer lt 7 For more information see Ess 1995 7 4 2 Gaussian split Ewald Gaussian split Ewald computations are configured as shown in force nonbonded far type gse common options sigma_s Os r_spread Repread k GSE spreads the point charges by a real space convolution with a Gaussian of width o sampling the result on the mesh The mesh charges are spectrally convolved with a kernel by means of the Fourier transform The forces and energies are then accumulated using another real space convolution by a Gaussian of width s The spectral convolution kernel is a Gaussian of width k 20 which is a Gaussian of width deconvolved twice by a Gaussia
44. applicable configuration file sections 3 Most simulations now require that the energy of the system be equilibrated so that initial forces between atoms are small One way to do this is to minimize the potential energy of the system Desmond provides two means of doing this The first is by Brownian motion through the use of the brownie_NVT or brownie_NPT inte grators or by gradient minimization through the minimize application You may not need to use equilibrated if your system was prepared with care to avoid energetic strains or if it has already been equilibrated with another tool On the other hand depending on how the structure file was obtained you may wish to use minimize even if you don t intend to run mdsin in order to rectify strange conformations resulting from the homology model or undesired artifacts resulting from x ray crystallography To minimize the energy of the system the structure file and associated configuration file are input to minimize which changes the atom positions slightly as needed It then outputs another structure file but does not change the configuration file 4 The new structure and the configuration file are now input to mdsim which executes the simulation possibly for days or weeks writing the results as frame files at the configured intervals of simulated time Analyze the results 3 8 Using Desmond 13 Desmond Users Guide Release 3 4 0 0 7 5 The frameset and configuration f
45. argument Since there are no integers in the interpreter a length one array should be used instead The element of this array will be rounded to get an integer and the differential of the germ will be discarded If a function requires a particle identifier then this should be a reference to a particle obtained by the atomsel function 11 3 3 Static Variables The interpreter has the ability to retain the value of certain variables for use on later time steps The variables that should be preserved for future time steps are declared in the header with the static keyword The type array length of each static variable must be specified in parentheses after the variable name Static variables can be read like any other variable but storing values in static variables must be done with the store function The first argument to the store function is the variable name while the second argument to the function is the value to be stored It is important to note that the action of store is delayed and the values of static variables do not change until the end of the time step For this reason all references to a static variable on the same time step will give the same value regardless of stores executed on that time step By the next time step any stores will have had their effect and changed the value of the static variable If a variable is referenced before a value has been stored in it then the value of the variable will be a zero array of the correct l
46. axes therefore set 21 n2 ng to 0 0 0 Desmond can nearly always determine an efficient global cell partitioning When the global cell isn t a rectangular volume Desmond issues a warning For example a hexagonal prism has X and Y vectors of the same length but the Z axis could be any length In this case if you set n na ng to 0 0 0 Desmond generates a partition but prints Automatic partitioning is untested for global cells with off diagonal boxes If you see this warning check the partitioning to ensure that 1t meets the criteria discussed above The margin A is a user provided upper bound on the maximum distance any particle will move between migration steps events Certain data structures within Desmond such as pairlists will rely on A to be faithful To tune various internal parameters for best performance Desmond needs an estimate of particle density est_pdens per home box By default Desmond sets d by computing an average density from the structure file For most simu lations it s safe to omit est_pdens in which case Desmond uses its default However if the density of particles in the system is highly inhomogeneous set d to e the maximum number of particles that could exist in a single home box e multiplied by the number of home boxes e divided by the volume of the global cell Configuring the global cell is summarized in Table 5 1 Configuration for global cell name description
47. bonds set of bonded neighbor atoms Atom chain chain name Atom charge charge in e Atom insertion PDB insertion value Atom mass mass in AMU Atom name atom name Atom occupancy PDB occupancy Atom radius a vdw radius value Atom resid PDB residue id Atom resname residue name Atom segid segment name Atom type VMD atom type Atom addbond atom add bond between self and atom Atom delbond atom remove bond between self and atom Frame box unit cell vectors as ROWS of 3x3 matrix Frame pos positions as rows of Nx3 matrix Frame time physical time Frame vel velocities as rows of Nx3 matrix Frame __init__ natoms new Frame with given number of atoms Frame moveby x y z shift the positions by the given amount Frame select inds new Frame with selected atoms Reader atoms copy of the atoms in the structure Reader natoms number of atoms Reader nframes number of frames 1 if not known Reader topology bond neighbor list Reader frame i Frame at index 1 Reader frames iterator over frames Writer natoms number of atoms in output file Writer path path of output file Writer close close the writer Writer frame f write Frame f Plugin can_read can create a Reader Plugin can_write can create a Writer Plugin name name of the plugin Plugin prettyname pretty name Plugin version major minor version
48. controlled by two parameters a relative tolerance 6 and a maximum iteration count m Iteration ceases if each particle pair distance is within a factor of 1 6 of its constrained distance A value of 1078 is suitable for most simulations The convergence rate is high enough that usually fewer than five steps are needed In the event that the constraint iteration fails Desmond prints a warning to the simulation log Regardless of the precision single or double used for the atomic coordinates the M SHAKE implementation per forms its calculations in double precision If the atomic coordinates are in single precision some error is inevitably introduced when these M SHAKE results are converted to atomic coordinates which could in principle be recovered at the next M SHAKE update This cumulative error is recovered by employing a novel algorithm we call reshake at the cost of additional arithmetic An alternative constraint algorithm is used for water constraints since the constrained molecule is a rigid body This algorithm due to Reich Rei 1994 derives a fixed rigid motion approximation to the constrained motion generally needing fewer arithmetic operations to preserve constraints to full precision Table 8 3 Configuration for constraint name description exclude constraint terms to turn off Optional by default empty List of names include constraint terms which must be turned on overrides exclude Opt
49. dimensional vectors such as r the positions of n particles are indexed differently The manual does not use r to refer to one of its 3n components but instead r denotes the i three dimensional vector in r which is the position of the it particle in this case 6 Chapter 2 Preface CHAPTER THREE KEY CONCEPTS This chapter explains the basic ideas underlying Desmond and describes how Desmond fits into a workflow 3 1 What is Desmond Desmond is a suite of computer programs for carrying out molecular dynamics simulations Such simulations model the motion of a collection of atoms a chemical system over time according to the laws of classical physics A collection of atoms representing such real world components as a protein molecule in water undergoing a structural change or a drug molecule interacting with a protein Desmond models solvents such as water explicitly as individual water molecules The chemical system exists in a thermodynamic environment which represents the conditions under which the sim ulation is carried out This environment mimics the experimental conditions whether the temperature or pressure is regulated for example or whether the system is isolated so that it cannot exchange energy with its environment The chemical system occupies a three dimensional volume of space of a specified size and each atom is generally represented by a particle at a specific position in that space Motion is simu
50. evaluate all the different forces between particles a program needs to be able to find them within a DMS file that may well contain any number of other auxiliary tables The DMS format solves this problem by providing a set of metatables containing the names of force terms required by the forcefield as well as the names of the tables in which the force term data is found The force terms are placed into one of four categories bonded terms constraints virtual sites polar terms Metatables for local particle interactions shows the names and descriptions of those tables The first four tables all of which refer to local particle interactions have the same schema shown in Local interaction metatables Each row in any of these four metatables corresponds to a unique functional form documented in later sections 17 2 Forcefields 129 Desmond Users Guide Release 3 4 0 0 7 Table 17 6 Metatables for local particle interactions metatable name description bond_term Interactions representing bonds between atoms including stretch angle and dihedral terms as well as 1 4 pairs and position restraints constraint_term Constraints on bonds and or angles involving a reduction in the number of degrees of freedom of the system virtual_term Similar to a constraint a set of parameters describing how a pseudoparticle is to be positioned relative to a set of parent atoms polar_term Similar to a virtual site a set of para
51. in the network of replicas The configuration for remd gragph is identical to that of remd except that the type and cfg sections are replaced by a new section called graph remd graph graph edges type linear nodes T1 T2 T3 type all to all nodes T1 T4 T5 J w ll J J N ll Hs e e T4 22 Chapter 4 Running Desmond Desmond Users Guide Release 3 4 0 0 7 T5 remd graph deltaE optional section first tf interval ti name p The graph section of the remd graph config must contain an edges section which is a list of edge declarations Each edge declaration has two fields nodes which is a list of symbolic replica names and t ype which specifies how those replicas are connected In an edge declaration of type linear edges are created between the nodes that are neighbors in the corresponding nodes list for type all to all edges are created between all nodes in the declaration The set of all edges is the union of the edges in all the declarations In our example we have edges T1 T2 and T2 T3 coming from the first declaration and edges T1 T4 T1 T5 and T4 T5 coming from the second declaration for a total of five edges The number of replicas in the simulation is given by the number of unique node names in the edges declarations For each name the graph section may also contain config overrides keyed to the name of the replica Once the set of ed
52. research tool The SOFTWARE is provided as is For the avoidance of doubt DESRES and its licensors shall have no maintenance upgrade or support obligations with respect to the SOFTWARE DESRES ITS AFFILIATES AND ITS LICENSORS DISCLAIM ALL WARRANTIES EXPRESS OR IMPLIED INCLUDING WITHOUT LIMITATION ANY WARRANTIES OF ERCHANTABILITY FITNESS FOR A PARTICULAR PURPOSE TITLE AND NON INFRINGEMENT OR THAT THE SOFTWARE WILL OPERATE UNINTERRUPTE OR ERROR FREE OR MEET LICENSEE S PARTICULAR R H P ab I D EQUIREMENTS LICENSEE AGREES THAT DESRES AND ITS AFFILIATES SHALL NOT BE ELD LIABLE FOR ANY DIRECT INDIRECT SPECIAL CONSEQUENTIAL EXEMPLARY UNITIVE OR INCIDENTAL DAMAGES WITH RESPECT TO ANY CLAIM BY LICENSEE OR ANY HIRD PARTY ARISING OUT OF OR RELATING TO THIS AGREEMENT OR USE OF THE SOFTWARE OR ANY DERIVATIVE WORK BASED ON THE SOFTWARE 6 Ownership Rights LICENSEE acknowledges that the SOFTWARE is the sole and exclusive property of and is valuable confidential and proprietary to DESRES and its licensors including without limitation all rights to patents copyrights trademarks trade secrets and any other intellectual propert
53. that are chemically bound Bonded forces are of at least three kinds Ore 0 a MA Ay Ay Figure 3 1 The various kinds of bonded forces e stretch Depends on the distance between the centers of two atoms sharing a bond e bend Depends on the angle between two bonds shared by one atom with two other atoms e torsion Depends on the torsion angle between two planes each defined by a group of three atoms where two of the atoms are shared between the groups A normal torsion is defined by a sequentially connected set of four atoms and an improper torsion has a more general relationship among its atoms In addition some force fields define other bonded terms Nonbonded force is the sum of two forces electrostatic and van der Waals Both kinds of nonbonded forces are a function of the distance between the two atoms In principle electrostatic and van der Waals forces must be computed between every pair of atoms in the system In practice however the magnitude of van der Waals forces falls off rapidly with distance becoming negligible between pairs of atoms separated by more than a certain distance referred to as the cutoff radius Therefore the simulation can restrict van der Waals calculations to only nearby atoms thus improving performance by reducing the number of computations Desmond must perform The cutoff radius cannot be used to limit electrostatic interactions however without seriously compromising accuracy Instead
54. the m_x_coord attribute takes on the values of 0 326 and 0 431 in the first and second m_at om records respectively Note Two kinds of ct blocks exist full or partial indicated by the respective name components f_ and p_ Partial blocks contain only attributes and values that override the corresponding values in the preceding full block Desmond makes no use of this feature You can think of each ct block as containing a complete description of a chemical system and the interaction between its particles Desmond reads all the ct blocks in a structure file and simulates them together in one chemical system with two exceptions e ct blocks with the attribute ffio_ct_type equal to full_systen are not included in the simulation Such ct blocks are used by Maestro for visualization e ct blocks corresponding to alchemical stages are combined into a single block before being loaded into the simulation More about preparing structure files for alchemical simulations can be found in Preparing the structure file for Free Energy Simulations 18 1 1 Global cell Desmond carries out simulations in a three dimensional region of space called the global cell described in Space The dimensions of the global cell are specified by the three shift vectors a b and c which together determine the shape of the global cell These shift vectors are specified in the ct attributes given in Global cell specification Table 18 1 Global cell specification
55. the next section 11 5 Examples The following sections give examples of enhanced sampling configurations to illustrate the uses of the enhanced sampling plugin 11 5 1 Center of mass restraint This example shows the use of an umbrella potential to harmonically restrain the center of mass for a group of particles In this example the masses of all particles are assumed to be the same declare_output name cvseq first 0 0 interval 0 1 spring 1 0 center array 4 0 5 0 6 0 p atomsel index 21 22 23 25 26 29 sum_val series i 0 length p diff pos p i center norm2 min_image diff y disp2 sum_val length p print sqr_disp disp2 spring disp2 11 5 2 Metadynamics for a dihedral angle This example demonstrates the use of metadynamics on dihedral angles In this case the sine and cosine of the angle are biased to avoid the derivative singularities associated with inverse trigonometry Biasing angles based on sine and cosine can be understood in the following way For a Gaussian centered at sin and cos with width w we have a sin sin g _ cos or 0 0 l 2w2 2w2 w2 In the case that the width is small this function is approximately a Gaussian in the angles with width w This function differs only by normalization from the normal distribution on the circle also known as the von Mises distribution 106 Chapter 11 Enhance
56. this print statement and the value of a is sent to the output The output file and frequency is controlled by the name first and interval parameters specified in the declare_output header The output side effect occurs only on the rank 0 process rad_gyration Class Normal Arguments e gids list of particles 1 2 2 A OEE Y a Return gt ig ri rj I Where r and r are the positions of particles i and j from the given list Returns the radius of gyration as defined in the wikipadia article of the same name under the molecular applications section rmsd Class Normal Arguments e model array of the model coordinates This argument should be of the form x1 y 21 2 Ya and the length of the array must be three times the length of position_gids e position_gids array of particles e weights array of the length as the position_gids array This argument is optional and if omitted all particles have the same weight The gradient of the RMSD with respect to weights is ignored 151 Desmond Users Guide Release 3 4 0 0 7 Return Minimum RMS distance between the positions described by position_gids and the structure described by model The minimum is taken of all possible affine transformations of the model To compute RMSD with periodic image ambiguities the following convention is used The minimum image displace ment of each GID is calculated with respect to the previous GID in the position_gids a
57. to skip You may use multiple match options together So to select only the potential and kinetic fields of a frameset run fsdump match POT_ENERGY match KIN_ENERGY foobar dtr The max option is used to trim very long output fields if you simply want to see a truncated view of a field So for example max 12 will allow you to see the first 3 position triples By default floats and doubles are printed in decimal using default formats that while they use a sufficient number of digits can not precisely represent all the bits of precision stored internally in the frame Using the hexfloat option will print the floating point values in a hex format that while not easily readable does perfectly represent all bits of precision in the double and float values The json option creates json Javascript object notation compatible output that can be fed into any standard json reader While slightly less readable json output is easier to machine parse 13 3 2 framedump The framedump command works just like sdump but works on a single frame file The command can be used to examine the common fields in the metadata frame file for instance framedump begin n nd n match xxx matchnot xxx max n hexfloat json framefile framefil 13 3 3 fstime fstime lists the number of frames and the last time contained in a frameset directory S fstime framesetdir 105
58. users can select this version by replacing their alchemical_pair_12_6_es terms with identically parameterized alchemical_pair_softcore_es terms Although the path of changing V from Vo to V should not in theory affect the outcome of the free energy calculation in practice the choice of A path determines the precision of calculated AF as well as the stability of the simulations For instance if an atom has different Lennard Jones parameters in states A and B at intermediate va it is interacting with other atoms through the soft core potential Unlike the Lennard Jones potential that rises steeply to infinity as the inter atomic distance r decreases to zero the soft core potential remains bounded for r 0 This means that other atoms can be infinitesimally close to this atom If the concerned atom has a nonzero partial charge infinite electrostatic energy results therefore it s important to turn off the partial charges on mutating atoms before changing their Lennard Jones interactions Here is a sensible schedule of alchemical transformation weights bondA 1 00 1 00 1 00 0 75 0 50 0 50 0 50 0 25 0 00 0 00 0 00 bondB 0 00 0 00 0 00 0 25 0 50 0 50 0 50 0 75 1 00 1 00 1 00 qA 1 00 0 75 0 50 0 25 0 00 0 00 0 00 0 00 0 00 0 00 0 00 qB 0 00 0 00 0 00 0 00 0 00 0 00 0 00 0 25 0 50 0 75 1 00 vdwA 1 00 1 00 1 00 1 00 0 75 0 50 0 25 0 00 0 00 0 00 0 00 vdwB 0 00 0 00 0 00 0 00 0 25 0 50 0 75 1 00 1 00 1 00 1 00
59. where N is a unique integer identifier for a particular table multiple cmap terms in torsiontorsion_cmap can refer to a single cmapN block The format of the cmap tables consists of two torsion angles in degrees and an associated energy cmap tables must begin with both torsion angles equal to 180 0 and increase fastest in the second torsion angle The grid spacing must be uniform within each torsion coordinate but can be different from the grid spacing in other torsion coordinates More information can be found in Bro 2004 Table 7 6 Schema for each of the tables holding the 2D cmap grids name type description phi FLOAT phi coordinate DEGREES psi FLOAT psi coordinate DEGREES energy FLOAT energy value ENERGY The CHARMM27 forcefield uses six cmap tables which have names cmap1 cmap2 cmap6 in DMS 7 2 Bonded pair and excluded interactions 53 Desmond Users Guide Release 3 4 0 0 7 Table 7 7 Schema for the torsiontorsion_cmap table name type description cmap INTEGER name of cmap table po INTEGER Ist particle pl INTEGER 2nd particle p2 INTEGER 3rd particle p3 INTEGER 4th particle p4 INTEGER 5th particle p5 INTEGER 6th particle po INTEGER 7th particle p7 INTEGER 8th particle These terms are in the cmap Hamiltonian category 7 2 6 Position restraint terms Particles can be restrained to a given glob
60. 4 0 0 7 minimize requires an integrator section even though all parameters in that section are ignored during the calculation We recommend that you use the same configuration for minimization and dynamics appending the minimize section to the mdsim configuration discussed above minimize handles constraints differently from mds im for a discussion see Adding constraints Table 4 6 Configuration for minimize name description plugin See Using plugins configuration migrate_interval Number of minimization steps between each migration event Optional by default 1 Integer gt 0 m Number of state vectors to use during L BFGS minimization Optional by default 3 Integer gt 0 maxsteps Maximum number of steps to iterate Optional by default 200 Integer tol Stopping tolerance for gradient norm Optional by default 1 0 Energy Length gt 0 stepsize Norm of first step Optional by default 0 005 Length gt 0 switch Minimum gradient before switching to L BFGS Optional by default 100 0 Energy Length gt 0 sdsteps Minimum number of initial steepest descent steps Optional by default 0 Integer dt A fake time scale for the minimize step Optional by default 1 0 time gt 07 4 4 5 vrun The vrun application is used to analyze structure files and trajectories It loads successive trajectory frames written by mds im or per replica frames written b
61. BAR method see Bennett acceptance ratio method to compute AF because of its lower statistical variance The variance in the computed AF is small only when the two Hamiltonians are similar such that the two systems overlap significantly in phase space In order to compute AF when Ho and H are very different we introduce n 1 interpolating Hamiltonians H where A i n 0 lt i lt n between Ho and H4 such that each pair of adjacent Hamiltonians is similar enough that the corresponding systems overlap significantly in phase space This family of Hamiltonians therefore provides a smooth and gradual transition from the initial state Ho to the final state H4 To compute the free energy difference between Hy and H n independent simulations are run for each A i n Each such simulation computes a pair of energy differences wet ww where ww Hit 41 n Hi n and ww FA 1 n Ai n sampled at a prescribed time interval t The free energy differences be tween the associated consecutive pair of H is then estimated from the wet we samples using the Ben nett acceptance ratio method These estimates W w are written to the output file by the name specified in force gibbs output name in the format shown in w tp WES wae t WED wet tet mt WES whee Combining two outputs from simulation iw and iw 1 we can estimate the free energy difference AF Imot in between systems H n and Hs 1
62. Desmond Users Guide Release 3 4 0 0 7 D E Shaw Research December 18 2012 CONTENTS Front matter 3 A AS AAN 3 1 2 BGOPYRIGNE eiir eee A EAS DE A e PRS 3 1 3 Trademarks sns zs ro a a e a ds 3 Preface 5 21 Intended audience lt lt a sude era Shoe ae eka ad dea ee wa Seed eee dw ees 5 2 2 Prereguisites cuio a a A GS ae aE SES a da Re Sd 5 2 3 Format conventions e be ee eda cd A A e eee eae bea eo 5 24 About theeguatio s s sorea meca e aep aS Ae ea ha e a we eS ap a 6 Key Concepts 7 3 11 Whatis Desmond se 2 0 Lan ed Eee eee e a ada rd 7 3 2 FOMES pie Aa ed A A A 8 8 33 Particles e sera ges we oy ee nE e a eS A ey ak A Ge Gace ee ws ag a 9 SA Porceshelds s s scra ow a BS AR Oe ae ee a a ee al Go ae Y 9 3 3 E O SB AS ALR gos ae aL G 10 30 ig Ch a o be bA Sa ea GRAY ae eA wk ee Se ae ea A 10 37 INE 11 3 05 Using Desmond wand os kee a YA ee ea A eet eee GS ee AA ees 12 Running Desmond 15 4 1 Aboutconhipuration 4 64 5 3445846480 ed awe Re eee eed ee ea amp 15 4 2 Invoking Desmond sorie e eona a Be A ae BS ds Bee a ley Bees 8 16 4 3 Running Desmond in parallel s ss 2 2 ere ew ee ee ee eee E 19 44 Configuring Desmond applications s s e s sos siea ee e opte a a SS 20 4 5 Naming output fil s lt s peceni ce spade A e A E a e 25 4 6 Configuring the built in plugins oeo aop eok ee ee 25 47 Comigurine Optional sections sese setga a a e 37 The Global Cell 39 Sel Paralleizaon 20d rr ed i et eae obs
63. E files require special attention Before running mae 2dms run the prep_alchemical_mae pro gram on the MAE file This program interprets the fepio_fep sections of the MAE file and converts the MAE file to a form more amenable to conversion to DMS format Once you have successfully converted a forcefield containing MAE file to DMS you are ready to begin equilibration and minimization 6 2 Preparing a Desmond DMS file Preparing a DMS file from scratch can be divided into four main steps First a DMS file must be constructed that contains all the atoms and bonds in the structure including ions waters protons etc along with a specification of the global cell Second this DMS file serves as the input to Viparr which adds forcefield information Third the build_constraints program is used to constrain bonds between hydrogen and heavy atoms Finally additional atom properties may be specified in order to perform specialized tasks such as energy group analysis or biasing force application 6 2 1 Constructing an input DMS file for Viparr The simplest method for preparing an input DMS file for Viparr is to use VMD VMD provides a number of tools for building structures in psf pdb format Once you have a molecule in VMD containing the full set of atoms and bonds you can write out the structure in DMS format using the dms file plugin Alternatively a DMS file may be produced by any tool that can write to the SQLite file format The input DMS f
64. ELOCITY FLT 3 natoms in Angstr ms picosecond UNITCELL FLT 9 Unit cell shift vectors as Ax Bx Cx Ay By Cy Az Bz Cz 13 2 Soft catenation option Multiple frameset directories can be soft catenated by listing the directory pathnames in a STK file ess tee kay file name suffix stk file separated by newlines Tools like the Python frameset tools see below the VMD trajectory reader and mol file can read STK files anywhere a DESRES trajectory file DTR file name suffix dtr is expected 13 3 Command line tools for framesets Frameset files have internal binary structure and are difficult to interpret manually The frameset library includes some programs that allow users to inspect view and correct framesets 13 3 1 fsdump f sdump is used to look at the times fields and data contained in every frame in a frameset Command line options control begin end frames which fields are viewed and the maximum number of items in each field to view 112 Chapter 13 Trajectory Format and Analysis Desmond Users Guide Release 3 4 0 0 7 fsdump begin n nd n match xxx matchnot xxx max n hexfloat json framesetdir framesetdir The begin option defaults to frame 0 end defaults to 1 negative indices count from the back so the 1th frame is the last frame 2nd is second to last frame etc The match and matchnot options signify fields to pick or fields
65. H_NVT List of Langevin parameters 9 6 9 Ber_NPT Berendsen constant temperature and pressure The Ber_NPT dynamical system is configured as shown in integrator Ber_NPT barostat tau Tp kappa Kk min_contraction_per_step sP in max_expansion_per_step Shax 84 Chapter 9 Dynamics Desmond Users Guide Release 3 4 0 0 7 tau T min_velocity_scaling Smin max_velocity_scaling Smax Just as a Berendsen NVT simulation does not sample according to canonical distributions a Berendsen NPT simula tion does not sample according to probability distributions in isothermal isobaric equilibrium ensemble It employs feedback control systems which try to keep the instantaneous temperature and pressure close to their reference values Temperature control is carried out similarly to Ber_NVT we discuss only the pressure control here Pressure control is performed by scaling the dimensions of the cell at each full timestep Ay If the instantaneous scalar pressure P Tr P 3 is greater than the target pressure Po the cell is expanded to release the extra pressure if P lt Pa the cell is contracted to build up pressure The scaling is done gradually according to a given parameter K which estimates of the compressibility of the system 1 O B ASS B OP and a relaxation time Tp A A B 1 a B In the isotropic case this is achieved by scaling each axis of the global cell
66. ID is given by gid pos_inner_prod Class Normal Arguments e gids array of gids e weights array same length as gids 150 Chapter 19 Enhanced sampling function reference Desmond Users Guide Release 3 4 0 0 7 Return 5 weights x pos gids after periodic image correction This function is useful for computing center of mass center of geometry and dipole moments To compute the inner product with periodic image ambiguities the following convention is used The minimum image displacement of each GID is calculated with respect to the previous GID in the gids array The location of a particle for the purposes of the inner product is then the sum of all these minimum image displacements for each adjacent pair in the gid array going back to the first particle The user must guarantee that each adjacent pair in the position_gid array is less than 1 2 of the unit cell apart If this condition is violated then particles may be wrapped to the wrong side of the cell distorting the inner product pow Class Normal Arguments e a array of positive numbers e b array of numbers same length or length 1 Return a performed elementwise or threaded in b If a is not positive the result is undefined print Class Special Form Arguments e printname string a array Return returns its argument a print is used to log values from the interpreter to a file printname is used to control the name associated with output from
67. LUDING NEGLIGENCE OR OTHERWISE ARISING IN ANY WAY O OF THIS SOFTWARE EVEN IF ADVISED OF THE POSSIBILITY OF 20 2 5 PCRE license PCRE is a library of functions to support regular expres and semantics are as close as possible to those of the P El Release 7 of PCRE is distributed under the terms of the specified below The documentation for PCRE supplied in directory is distributed under the same terms as the so The basic library functions are written in C and are fre included in the distribution is a set of C wrapper fun THE BASIC LIBRARY FUNCTIONS Written by Philip Hazel Email local part ph10 Email domain cam ac uk University of Cambridge Computing Service Cambridge England Copyright c 1997 2007 University of Cambridge All rights reserved THE C WRAPPER FUNCTIONS Contributed by Google Inc Copyright c 2007 Google Inc All rights reserved TY OR TORT UT OF THE USE SUCH DAMAGE El sions whose syntax erl 5 language BSD licence as the doc ftware itself estanding Also ctions 160 Chapter 20 Licenses and Third Party Software Desmond Users Guide Release 3 4 0 0 7 THE BSD LICENCE Redistribution and use in source and binary forms with or without modification are permitted provid
68. NSEE shall discontinue all use of the SOFTWARE and shall either a promptly return all copies of the SOFTWARE and any RELATED MATERIALS to DESRES or b subject to DESRES s prior consent provide DESRES with a certificate of destruction of all copies of the SOFTWARE and any RELATED MATERIALS Notwithstanding the foregoing only Paragraph 1 of this Agreement shall not survive the termination of this Agreement GI 8 Government Use The SOFTWARE and the accompanying documentation are commercial items as that term is defined in 48 C F R 2 101 consisting of commercial computer software and commercial computer software documentation as such terms are used in 48 C F R 12 212 Consistent with 48 C F R 12 212 and 48 C F R 227 7202 1 through 227 7202 4 if LICENSEE hereunder is the U S Government or any agency or department thereof th SOFTWARE and the documentation are licensed hereunder i only as commercial items and ii with only those rights as granted to all other end users 20 1 Licensing Desmond for Non Commercial Research 157 Desmond Users Guide Release 3 4 0 0 7 pursuant to the terms and conditions hereof 9 General This Agreement and its enforcement shall be governed by and construed in accordance with the laws of the State of New York wi
69. PEPE e ER 39 5 2 CONASUTAION sa a ri e OA A we eb SARE oe oa ada SS 42 5 3 Migration s e se ee a oe Ga a ee eae ea a Oe Gee Go RO oe ee ae a e Y 43 Preparing a structure file 45 6 1 Converting a Desmond 2 0 2 2 str ct re file x s s 24464 240 e008 444 24 5 YES SERS 45 6 2 Preparing a Desmond DMS fil 2 0 re 444 cae He 6G de Soe ae Be RO ee ee 46 Calculating Force and Energy 49 10 11 12 13 14 15 16 17 18 19 20 Ll Configuring force elds s s esii e ld He go A A hes As Bee a 7 2 Bonded pair and excluded interactions e ioe ee a we a e 7 3 Van der Waals and electrostatic interactions e 14 Nonbonded tar mteractions s pes 5 hese ee ea SA ew ae EO RE ee eae Rees Constraints 8 1 Single precision resolution and constraints 2 2 2 ee ee ee ee Dynamics Ol Particles and Mechanics ore agia ee ig Gwe BP a BM pd wee OB ack ae A Se ia 92 Integrator ooo hb6 ei hee a Boe Geb ePbdd de Phe wee eee Sede ea A 93e RESPAL 4 64 2 443 4 263 4 e484 ehh Set A A Se ee ee e ER OA PIESSUTEs noes WA RR ee YO ee HK EA Soe Ee a se BAe a OD Temperate 2046 24 4ec ed ee ee We Yes ar Re ERO ED we A Se eee ge PAR SS 96 Dynamicalisystems i dna be cop a ee eb da bet gs Sea eee ee eb ee eee Bs Free Energy Simulations 10 1 Configuring free energy simulations 2 44 05 445040508 eee ee ae E Enhanced Sampling and Umbrella Sampling Li Introduction lt edi i i55 62645 4 255 be bbe 2 b
70. Waals and electrostatic interactions The nonbonded forces are configured as shown in force nonbonded n_zone nz sigma o r_cut Reut near type near type far type far type Van der Waals interactions decay rapidly with distance whereas electrostatic interactions are split into a rapidly de caying part near and a slowly decaying part far Near nonbonded interactions are computed by summing them over all pairs except the excluded ones within a distance Reut of each other Far electrostatic interactions are computed by an Ewald summation technique specified in the far configuration see Nonbonded far interactions and a sum over certain designated electrostatic correction terms 7 3 Van der Waals and electrostatic interactions 57 Desmond Users Guide Release 3 4 0 0 7 The electrostatic potential is decomposed as Ndj quajerte r v 20 r qiqjerf rij vV 20 rij 1 where q and q are the charges of particles and j and erf r and erfc r are the error function and the complementary error function respectively The term involving erfc falls off quickly with distance it is usually computed by an interpolating function truncated to 0 for r gt Reut The erf term is essentially the far interaction The value of is typically chosen such that erfe r y2 0 is small at the cutoff radius Reut a common choice is o Reut 3W 2 which assumes erfc 3 2 209 x 107 is suff
71. _aj The second atom in the exclusion in orignal_ct Negative atom numbers can appear here by the same convention as in atommaps fepio_pairmaps This indexed block maps the pairs from original_ct onto perturbed_ct Table 18 11 fepio_pairmaps properties Property name Description i_fepio_ti Pair number in original_ct Negative exclusion number indicates a pair involving at least one dummy atom i_fepio_tj Pair number in perturbed_ct Negative exclusion number indicates a pair involving at least one dummy atom i_fepio_ai The first atom in the pair in original_ct Negative atom numbers can appear here by the same convention as in atommaps i_fepio_aj The second atom in the pair in orignal_ct Negative atom numbers can appear here by the same convention as in atommaps The next Example shows an excerpt from a structure file for an alchemical free energy simulation in which a methyl group in ethane vanishes and is replaced by another methyl group The first ct block describes the original ethane molecule and the second ct block describes the ethane molecule with one methyl group replaced by another albeit identical methyl group The second ct block contains the fepio_fep section that details the mapping of the second molecule onto the first one The third ct block describes the solvent in which the transformation takes place from the ethane in the first ct block to the ethane in the second
72. a plugin of that type has already been loaded if it has this plugin is not used the declaration is ignored In this manner the plugins of all extensions in the DESMOND_PLUGIN_PATH are loaded and examined Desmond then unloads any extensions who contributed no plugins and calls the boot method of each plugin declaration 2 Among other thing a plugin s boot method typically registers a concrete subclass of some interface class with an abstract factory under some name so that this subclass can be produced by the factory as directed by the configuration or the structure file 3 At some point in the parsing of the configuration file or the structure file a string identifying the subclass by its abstract type and registered name will direct an instance of the subclass to be created through the appropriate factory When Desmond shuts down the steps occur in reverse 1 Desmond calls the halt method as given in the plugin declaration for each booted plugin and 2 Desmond unloads the shared libraries 110 Chapter 12 Extending Desmond CHAPTER THIRTEEN TRAJECTORY FORMAT AND ANALYSIS Desmond writes time sampled data into trajectory collections These collections are stored in the file system and are called framesets These trajectories are a series of frames that represent snapshots of the simulation a various times Each frame has a collection of simulation data The data contains at a minimum information about chemical time t
73. al coordinate by means of the restraining potential V 2y 2 5 foul z0 feyly v0 fee 2 20 where fox fey fez are the force constants in Energy Length and zo Yo Zo are the desired global cell coordinates units of Length Terms in posre_harm are evaluated using this potential Table 7 8 Schema for the posre_harm table name type description fcx FLOAT X force constant in ENERGY LENGTH2 fcy FLOAT Y force constant in ENERGY LENGTH fcz FLOAT Z force constant in ENERGY LENGTH po INTEGER restrained particle x0 FLOAT x reference coordinate yO FLOAT y reference coordinate z0 FLOAT z reference coordinate These terms are in the posre Hamiltonian category 7 2 7 Exclusions Exclusion terms in exclusion are used to prevent calculation of certain non bonded interactions at short ranges The excluded interactions are typically those that involve particles separated by one or two bonds as these interactions are assumed to be adequately modeled by the stretch and angle terms described above Table 7 9 Schema for the exclusion ta ble name type description po INTEGER Ist particle pl INTEGER 2nd particle Desmond requires that p0 lt p1 for each term and every p0 p1 pair should be unique Exclusions are in the far_exclusion Hamiltonian category 54 Chapter 7 Calculating Force and Energy Desmond Users Guide Release 3 4 0 0 7
74. ame Columns in the tables have a name a datatype and several other attributes most importantly whether or not the column is the primary key for the table Rows in the tables hold a value for each of the columns Table names column names and datatypes are case preserving but case insensitive thus pArTiCLE is the same table as particle and NAME is the same column as name For more about the DMS format see Appendix DMS file format There are two main ways to prepare a DMS file for Desmond The first method is to convert an existing MAE file and all its forcefield data to DMS using the mae2dms tool described below The second method is to construct a DMS file containing just the minimal set of molecular structure information and build a forcefield for that structure using Viparr 6 1 Converting a Desmond 2 0 2 2 structure file If you already have an MAE file prepared either with viparr py from Desmond 2 0 2 2 or with Schrodinger s Maestro tool you can convert to to DMS using mae2dms mae2dms preserves all forcefield information including bonded terms vdw tables energy and temperature groups constraints cmap tables and position restraints Force field terms that were present and supported in Desmond 2 2 should be properly handled by mae2dms any forcefield type that was not present in Desmond 2 2 is not likely to be recognized and converted 45 Desmond Users Guide Release 3 4 0 0 7 Alchemical MA
75. an be added to the MAE file using versions of Viparr provided with Desmond 2 2 18 2 Preparing the structure file for Free Energy Simulations The sections below describe additional steps needed to prepare a structure file for ligand binding free energy simula tions and for alchemical free energy simulations 18 2 1 Ligand binding free energy simulations To prepare the structure file for ligand_binding free energy simulations specify the atoms that belong to the ligand To do so set the ffio_grp_ligand field in the m_atoms records to 1 for the ligand atoms and to 0 for other atoms The following Example shows an excerpt from a structure file for simulating the solvation free energy of methanol highlighting the fi0_grp_ligand field The first ct block describes the solute methanol having all the atoms in its fio_grp_ligand set to 1 The second ct block describes the solvent water having all the atoms in its ffio_grp_ligand set to 0 lines omitted E moot lines omitted s_ffio_ct_type lines omitted solute m_atom 6 i_m_mmod_type r_m_x_coord r_m_y_coord rm 2 coord s_m_pdb_atom_name i_m_atomic_number i_ffio_grp_ligand 13 0 683143 0 071748 0 090914 C1 6 1 2 16 0 463103 0 750632 0 140535 02 8 1 3 41 1 138147 0 383230 0 876254 H3 1 1 136 Chapter 18 Legacy Applications Preparing a Maestro structure file Desmond Users Guide Release 3 4 0 0 7 4 41 1 450629 0 486326 0 674057 H4
76. approximation is roughly given by ae o r _ CE 2 C 4 C 6 E 444 nr o r The relative error of db decreases with the fourth power of n and increases with the fourth power of C using inter polation to compute van der Waals potentials hence would require a much greater n Meanwhile increasing n increases the size of the table of spline coefficients as well as the size of the linear system numerically solved to set those coefficients The near nonbonded force calculation skips over excluded pairs if any Additionally for all excluded pairs i j a far exclusion calculation subtracts the contribution from the potential term q q erf ri V2 a frij and its associated force from the energy and the forces Like the near nonbonded terms this function is evaluated by a cutoff interpolation function Because the calculation is cut off for large r in practice the distance between excluded pairs of particles is always within a sensible Reut Table 7 17 Configuration for near name description taper Tapering strategy to use none shift clswitch c2switch r_tap Distance at which to begin the tapering strategy Length lt freut average_dispersion Correction factor for long range van der Waals interactions Optional by default calculated Length gt 0 7 3 Van der Waals and electrostatic interactions 59 Desmond Users Guide Release 3 4 0 0 7 Table 7 18 Sc
77. arostat step Integer gt 0 a multiple of the thermostat timesteps The particular parameters for the various kind of thermostat and barostat steps are discussed in sections on other integrators Stability of the PLS and Verlet inegrators Although the PLS NVE steps have accuracy advantages over Verlet NVE steps for integrating harmonic motion as well as advantages in reproducing certain thermodynamic statistics they can have decreased stability the maximum for which the simulation does not blow up in comparison Linear stability theory can be carried out analytically by integrating a harmonic oscillator and looking for modes which positive exponential growth but such analysis is not useful in making stability comparisons for schemes where the force field has been split into various components active in different phases any component may contain the hypo thetical harmonic potential Instead we have carried out an empirical analysis on a test system Sdhfr comparing the results of Verlet and PLS at different RESPA schedules The maximum 6 reported is the value of integrator dt for which the simulation judging by a sudden increase in energy drift began to become unstable Table 9 15 PLS stability comparison method schedule max 0 fs fraction of Verlet111 fraction of Verlet Verlet 1 1 1 3 7 1 00 1 00 Verlet 1 2 2 3 1 0 84 1 00 Verlet 2 2 2 2 0 0 54 1 00 Verlet 1 3 3 2 3
78. ates extensions files containing plugins by means of either of two environment variables DESMOND_PLUGIN_PATH and DESRES_PLUGIN_PATH You can specify more than one path to search for plugins by separating them with colons as in DESMOND_PLUGIN_PATH this is the first path this is the second The line above specifies two directories which are searched for plugins in the given order Many plugins are compiled with Desmond already and are therefore available to all its applications these are discussed in Configuring the built in plugins In addition you can implement your own plugins or use those developed by third parties Extending Desmond s functionality in this way is discussed in Extending Desmond Each application has a main loop consisting of one minimization or simulation step nds im remd and minimize or processing one trajectory frame vrun You can configure a plugin to run once at the beginning of a simulation or periodically at an interval of one or more steps Each application s plugin section of the configuration contains a 1ist under the key plugin that gives the names of main loop objects to create For example the plugins to call when the mdsim application runs appear in a list like the one below mdsim plugin list key key key1 type type keyn type type The key names appearing in the plugins list are arbitrary though for a given section they must be u
79. ation marks single or double quotation marks or backticks with internal quotes escaped with the character nil written as table An unordered set of key value pairs with distinct keys written keyl valuel key2 value2 keyN valueN A key is an alphanumeric unquoted string that can also include underscores list A sequence of values written valuel value2 valueN Syntax does not require that the values of a list be of similar type though for clarity we recommend following this convention 15 1 Examples Below is an example of a configuration file title this is an example an atom quoted string pi 3 14159 an atom bare string file myDoc txt an atom bare string matrix 1 0 0 0 1 0 0 0 1 a list of lists 122 Chapter 15 Appendix Configuration syntax Desmond Users Guide Release 3 4 0 0 7 options verbose yes Nsteps 100 vec 1 2 3 a table This configuration could be given to a Desmond application with either the include or c fg options as follows desmond include config_file or desmond cfg title this is an example pi 3 14159 file myDoc txt matrix 1 0 0 0 1 0 0 01 options verbose yes Nsteps 100 vec 1 2 3 The first of these reads a file named config_file which we assume holds the contents of the example The second gives the contents of the previous example as a string Configuration flags can be co
80. b in radians Result is in the range 0 7 This function is not safe to use if the angle is near 0 or 7 because the derivative of this function diverges It is preferable to use the function angle when possible array Class Normal Arguments e An arbitrary number of array arguments Return the concatenation of all arguments This function is useful for the creation of data arrays For example array 1 0 2 0 3 0 4 0 is a 4 element array because scalar literals are 1 element arrays atan2 Class Binary Threaded Arguments e y array e x array Return arctangent of y x computed according to the binary threading rules with the quadrants chosen according to the signs of x and y The range of this function is 7 7 The derivative of this function does not capture the discontinuity of the function at 7 If the angle crosses 7 there can be a discrete change in the potential without a corresponding derivative divergence This can cause energy drift in the simulation The user is advised to exercise caution if using atan2 in enhanced sampling potentials center_of_geometry Class Normal Arguments e gids array of gids Return center of geometry with periodic image handling To compute the center of geometry with periodic image ambiguities the following convention is used The minimum image displacement of each GID is calculated with respect to the previous GID in the gids array The location of a particle for the purp
81. be short enough to capture the vibrational frequency of the atoms you re modeling Yet the shorter the timestep the less simulated time you can compute in a practical period of clock time Desmond Users Guide Release 3 4 0 0 7 To enhance performance as much as possible Desmond implements a variety of features Some such as an algorithm used to minimize interprocessor communication are built into Desmond and require no action on your part Others require you to specify their use for example you can run Desmond in parallel using as many processes as your parallel environment can support Spreading the many computations among many processes can yield a significant increase in speed Still other performance features however don t make sense for every simulation therefore part of configuring a simulation is to set them as you require In order to make most effective use of Desmond then you ll need to learn certain details about the way it works Where relevant such performance issues are noted below and throughout the manual In addition to the simulations described above Desmond has the ability to perform Gibbs free energy simulations which compute the change in free energy of a chemical system as 1t evolves from one state to another These are described in detail in Free Energy Simulations 3 2 Forces The total force on a particle is the sum of bonded and nonbonded forces A bonded force is a force due to two or more atoms
82. ble 17 9 Schema for the nonbonded_combined_param table name type description paraml INTEGER 1st entry in nonbonded_paran table param2 INTEGER 2nd entry in nonbonded_param table coeffl FLOAT first combined coefficient other combined coefficients Only paraml and param2 are required the remaining columns provide the interaction dependent coefficients 17 3 Alchemical systems Methods for calculating relative free energies or energies of solvation using free energy perturbation FEP involve mutating one or more chemical entities from a reference state labeled A to a new state labeled B DMS treats FEP calculations as just another set of interactions with an extended functional form In order to permit multiple independent mutations to be carried out in the same simulation a moiety label is applied to each mutating particle and bonded term 17 3 1 Alchemical particles Any particle whose charge or nonbonded parameters changes in going from state A to state B is considered to be an alchemical particle and must have a moiety assigned to it The set of distinct moieties should begin at O and increase consecutively The set of alchemical particles if any should be provided in a table called alchemical_particle shown in Schema for the alchemical_particle table Table 17 10 Schema for the alchemical_particle table name type description po INTEGER alchemical particle id mo
83. ble with the multigrator 1 1 1 1 1 2 1 1 3 1 1 4 1 2 2 1 3 3 1 4 4 1 2 4 and 1 3 6 Note the none NVE type performs no position or momentum changes For every sequence of nr inner NVE steps a pair of thermostat steps are added to the beginning of the first and to the end of the last such that the full sequence is an NVT step spanning a chemical time of n76 Every ng inner NVE steps or ng nr NVT steps a pair of barostat steps are added to the beginning of the first and the end of the last such that the full sequence is an NPT step spanning a chemical time of ng The Verlet NVE type performs a standard RESPA integrator step splitting the force field into weighted components according to the schedule see RESPA The PLS is similar to Verlet in that it creates an integrator step from momentum and position updates similar to a RESPA step but the weights of the force components and individual time increments of each update have been somewhat modified such that true harmonic motions are approximated to higher order than x 97 The PLS steps are generally less stable than the analogous Ver let steps see below The timescales of the steps employed for thermostat and barostat updates are independent of ng nr and the RESPA schedule Each pair of steps of a given type updates its associated variables by an approximation to a differential equation evolved as described below for a total time equal to an inner timestep In the limit as 6 0
84. box since in this case the center of mass and the moment of inertia are ill defined There is a limit of 4 biasing potentials in the cm_moi list of the force BiasingForce configuration The user can monitor the action of the biasing potential from the data written at the specified time intervals to the output file The header in the output file reports the number of atoms in each cm_moi group the user should verify that these match the intended grouping Following this is a header line that labels each column of the subsequent data Each row of data corresponds to one moment in time beginning with the properties of the first cm_moi group followed by those of the ensuing groups The data reported are as follows e xci yci zci where i 0 1 the center of mass coordinates of the it cm_moi group in units of A e p1xi plyi p1zi the unit vector of the first principal axis of the it cm_moi group e p2xi p2yi p2zi p3xi p3yi p3zi the unit vectors of the second and third principal axes of the ith cm_moi group e Tli Bi Bi the diagonal moment of inertia tensor of the ith cm_moi group corresponding to the principal axes in the same order They are in units of amuxA From these data together with the parameters in the biasing potential configuration it is straightforward to compute the energy contributions from the biasing potential at each recorded moment 28 Chapter 4 Running Desmond Desmond Users Guide Release 3 4 0
85. by a factor s given by s 1 n P 1 3 Po P Tp More generally for non isotropic cases we scale B to B s where s satisfies a A s a 1 Tp PAP and the forms of a A and B are determined by the isotropy Note Berendsen is not expected to work with constant area isotropy To avoid changing the box dimensions too much in a single step each scaling factor s is constrained to Smin lt S lt Smax Table 9 12 Configuration for Ber_NPT name description barostat tau Relaxation time for Berendsen pressure control Time gt 0 barostat kappa Estimated compressibility of the system Pressure gt 0 barostat min_contraction_per_step Minimum factor for scaling the box in one timestep 0 lt Real lt 1 barostat max_expansion_per_step Maximum factor for scaling the box in one timestep 1 lt Real tau Relaxation time Time gt 0 min_velocity_scaling Minimum factor for scaling particle velocities in one timestep 0 lt Real lt 1 max_velocity_scaling Real maximum factor for scaling particle velocities in one timestep 1 lt Real 9 6 Dynamical systems 85 Desmond Users Guide Release 3 4 0 0 7 9 6 10 Brownian motion integrators ll A integrator brownie_NVT thermostat seed 7 delta_max Amax integrator brownie_NPT thermostat seed barostat thermostat seed iy tau Tp T_ref T
86. ce ratio method is implemented in the script bennett py It is tailored to work with output files of the above form 10 1 2 Binding free energy simulations force term keyx type binding weights es Co Cin see C vdw vo Vijn see v force nonbonded near type binding softcore same parameters as default force nonbonded far type binding pme binding gse same parameters as pme or gse Binding free energy simulations compute the free energy of adding a molecule called the ligand to the chemical system Effectively this free energy is the difference between e the system in which the ligand is fully interacting with the rest of the system and e the system in which the ligand is not interacting at all with the rest of the system Denoting the ligand degrees of freedom by rz and those of the rest of the system by rg the Hamiltonian of the system can be separated into three components H r Ay rv Hs rg V r rs where Hzr and Hs are the Hamiltonians of the ligand and the rest in isolation and V is the interaction potential between the particles of the ligand and the rest We introduce a family of interpolating Hamiltonians Hx r Ay rz Hs rg Vi rL rg such that Vo rz rg 0 and Vi rz rs V rL rg At present Desmond handles only the most common case where ligand molecules do not have covalent interactions with the rest of the system In terms of a classica
87. cle The functional form of the FBHW improper term is V fed where p o 0 where p do lt o d lt 0 where o lt o lt o o 0 where o lt do 56 Chapter 7 Calculating Force and Energy Desmond Users Guide Release 3 4 0 0 7 The improper dihedral angle phi is the angle between the plane ijk and jkl Thus fc is in ENERGY and phi0 is in RADIANS Table 7 14 Schema for the improper_fbhw table name type description fc FLOAT force constant in ENERGY RADIANS phi0 FLOAT equilibrium improper dihedral angle in DEGREES sigma FLOAT half width of flat bottomed region in DEGREES po INTEGER first particle pl INTEGER second particle p2 INTEGER third particle p3 INTEGER fourth particle The functional form of the FBHW posre term is V f 2d where dJ r r0 o where r r0 gt c 10 where r rO lt 0 This is not as general as the fully harmonic position restraint term in that you can t specify different force constants for the three coordinate axes Table 7 15 Schema for the posre_fbhw table name type description fe FLOAT force constant in ENERGY LENGTH x0 FLOAT equilibrium x coordinate in LENGTH yO FLOAT equilibrium y coordinate in LENGTH z0 FLOAT equilibrium z coordinate in LENGTH sigma FLOAT radius of flat bottomed region in LENGTH po INTEGER restrained particle 7 3 Van der
88. ction reference is given in Enhanced sampling function reference 100 Chapter 11 Enhanced Sampling and Umbrella Sampling Desmond Users Guide Release 3 4 0 0 7 11 3 1 Syntax Potentials are specified in the interpreter using an imperative syntax The program is divided into a header where global declarations are made and a body where executable statements are written Both the header and the body are semicolon separated lists of statements and the header is distinguished from the body only by the type of statements allowed in each section In the example configuration above the only header statement was the declare_output statement and the rest of the statements constituted the body Each statement in the body is either an assignment or an expression Each assignment is a variable name followed by an equal sign and then an expression Each variable may only be bound once i e this is a single assignment language and later references to the variable use the stored value of the variable The only exception to the single binding rule is 1f the variables are at different scopes as explained below Expressions are written in a style similar to C or Python Functions are called by writing the function name followed by a comma separated list of arguments enclosed by parentheses The binary operators x and are available and they obey the normal precedence rules Unary negation is indicated by writing a negative sign at the
89. ctions in this chapter will be written at this finer level of control and use y i to denote the element of temperature in which the i particle s temperature group occurs in other words x i j means particle is governed by thermostat j in temperature controlled simulations We set y i 0 when the group is not assigned a reference temperature Note Desmond prints a warning if some particles in the simulation have not been assigned a reference temperature Table 9 4 Configuration for temperature name description _ref The global reference temperature Temperature gt 0 T_groups i T_ref The reference temperature for thermostat i Optional defaults to the global reference temperature Temperature gt 0 T_groups i groups The temperature groups regulates by thermostat i List of Integers gt 0 9 6 Dynamical systems Three kinds of dynamical systems are available in Desmond 9 5 Temperature 73 Desmond Users Guide Release 3 4 0 0 7 e ordinary differential equations ODEs with certain energy and measure conserving properties e stochastic differential equations SDEs with invariant measures and e stochastic differential equations coupled to feedback control systems This section describes the supported systems in a mathematically exact and unconstrained form omitting the details of the integration method and the complexities of incorporating constraints A simulati
90. d x velocity in LENGTH TIME vy FLOAT Desmond y velocity in LENGTH TIME vZ FLOAT Desmond z velocity in LENGTH TIME nbtype INTEGER Desmond nonbonded type grp_temperature INTEGER Desmond temperature group grp_energy INTEGER Desmond energy group grp_ligand INTEGER Desmond ligand group grp_bias INTEGER Desmond force biasing group resid INTEGER Viparr residue number resname TEXT Viparr residue name chain TEXT Viparr chain identifier name TEXT Viparr atom name formal_charge FLOAT Viparr format particle charge occupancy FLOAT pdb occupancy value bfactor FLOAT pdb temperature factor Optional particle properties that may be added as additional columns in the particle table 17 2 Forcefields A description of a forcefield comprises the functional form of the interactions between particles in a chemical system the particles that interact with a given functional form and the parameters that govern a particular interaction At a higher level interactions can be described as being local or nonlocal Local particle interactions in DMS are those described by a fixed set of n body terms These include bonded terms virtual sites constraints and polar terms Nonlocal interactions in principle involve all particles in the system though in practice the potential is typically range limited These include van der Waals vdw interactions as well as electrostatics 17 2 1 Local particle interactions In order to
91. d Sampling and Umbrella Sampling Desmond Users Guide Release 3 4 0 0 7 declare_meta dimension 2 for sine and cosine cutoff 9 in units of widths first 0 0 begin dropping immediately interval 0 200 wait 0 2 picoseconds between drops name kerseq log kernels to kerseq initial no initial kernel file p atomsel index 14 15 16 17 height is 0 2 and widths are both 0 1 meta 0 array 0 2 0 1 0 1 dihedral_gid p 0 p i p 2 p 31 11 5 3 Well tempered metadynamics This example will use well tempered metadynamics to demonstrate metadynamics with dynamically varying heights For well tempered metadynamics the height of a Gaussian added at time t is given by hoe A where ho is the initial height V x is the metadynamics potential at the center position and T is a user specified temperature Since the metadynamics potential must be known before the Gaussian is added a small trick is used To evaluate the metadynamics potential without changing the potential metadynamics is called with a height of 0 0 In this case Gaussian kernels are added by this evaluation but they do not contribute to the potential They are however present in the kernel sequence file declare_meta dimension 1 cutoff 9 first 0 0 interval 0 200 name kersegq initial p atomsel index 0 1 h_0 0 020 initial height of gaussians w 0 1 width of gaussians kT1 0 6 sampling t
92. d as mpi must either be built in that is compiled as part of the Desmond executable or located in an extension specified by the path given in your DESMOND_PLUGIN_PATH environment variable destrier serial runs Desmond applications with a single process This gives you a means to check your code and find any other problems while your installation creates a usable parallel environment e destrier mpi uses the MPI destrier variant a common parallel programming specification imple mented as a library of C C or Fortran functions destrier other You can create your own destrier plugin by modifying the examples provided for the serial and mpi plugins Register the resulting plugin under the name of your choice supplying that name as the 4 3 Running Desmond in parallel 19 Desmond Users Guide Release 3 4 0 0 7 argument to the destrier parameter The parallel environment is initialized before checkpoint information is read Therefore if you re restoring from a checkpoint the dest rier flag must be set in the same way it was when you started the original simulation Note The mpi destrier plugin requires Open MPI version 1 4 3 or later If you wish to use a different parallel communication interface you ll need to compile your own plugin 4 4 Configuring Desmond applications The main Desmond applications are mdsim minimize remd and vrun as described in Applications and scripts Configurat
93. d constraint calculations For far electrostatic force calculations there are additional restrictions on the clone radius These restrictions are usually weaker than the above but are included for completeness In the case of PME A v2 where h is the Ewald mesh spacing in the i direction A is the margin discussed here and o is the PME interpolation order in the direction In the case of k GSE Retone 2 3 Y Olor 1 Ry oy 1 halos DY Rolone gt Rspread A 2 Va where Rspread is the k GSE spreading radius It is generally not necessary and is inefficient to choose the clone radius larger than what the above restrictions require There are also upper limits to the size of the clone radius These come from the parallelization of the global cell and particle image tracking which does not allow greater than nearest neighbor communications or certain kinds of 125 Desmond Users Guide Release 3 4 0 0 7 self overlapping clone regions First because Desmond communicates only with immediately adjacent boxes during migration the clone radius cannot be larger than the box dimension in any direction in other words Retone lt L where L is the home box dimension in the i direction This condition may restrict how many processes you can use to parallelize your chemical system At low levels of parallelism if a dimension 7 has been partitioned into only two boxes then we have the
94. d is the chemical time and the value of the enhanced sampling potential The user may specify additional output using the print function in symbolic expressions This allows the user to print the value of an arbitrary expression to aid in debugging and interpreting results Because of looping and other constructs in the symbolic expressions the amount of output generated may not be the same every time the interpreter is called In addition if print is called within a loop a large amount of output may be generated on each step For these reasons a structured output format is used instead of column output This output occurs only every interval picoseconds to a file whose name is given by the name parameter Each line of output represents one evaluation of the enhanced sampling plugin and is of the format namel value name2 value2 Each value is a list of floating point numbers 11 2 3 Example configuration The following is an example of a simple configuration that creates a harmonic potential between atoms with GIDs 10 and 20 declare_output name cvseq output file first 00 first output occurs at time 0 0 ps interval 0 020 output every 0 020 ps p atomsel index 10 20 select the needed particles 7 5 dist p 1 p 0 2 compute the potential The syntax of this m expression code is explained below 11 3 Interpreter This section documents the m expression syntax and semantics The complete fun
95. d pressure if any the integrator to use the length of the timestep the fineness of the grid to use for charge spreading how many processes to assign to a given dimension of the global cell and possibly many other such parameters By using different configuration files with the same structure file you can run different simulations 3 8 2 Applications and scripts Desmond consists of three main applications and several companion Python scripts mdsim The application that performs the molecular dynamics simulation minimize The application that prepares the molecular dynamics simulation if necessary by minimizing ener getic strains in the system so that they don t destabilize the simulation at the first few steps vrun The application used to analyze framesets output by mdsim Viparr The Python script that adds force field information to the structure file build_constraints The Python script that adds constraint information to the structure file 3 8 3 Output Timestep by timestep an atom traces a path through the global cell as the simulation advances The path that molecules take through the global cell is the trajectory Trajectories are writ ten out in a set of files representing a time series like the frames of a movie Each frame is a file containing the positions and velocities of all the particles and pseudoparticles in the chemical system at that particular timestep In addition to particle positions and velocitie
96. d velocities using the attributes listed in Table tab ppv and e a set of integer valued properties ff io_grp_name Desmond makes use of energy temperature cm_moi ligand and frozen groups described in Particles Table 18 2 Initial particle position and velocity specification particle property m_atom attribute ffio pseudo attribute X position m_x_coord ffio_x_ coord Y position m_y_coord ffio_y_coord Z position m_z_coord ffio_z_coord X velocity ffio_x vel ffio_x vel Y velocity ffio_y vel ffio_y_vel Z velocity ffio_z vel ffio_z vel Particles are loaded into Desmond in the order in which they appear in the structure file Within a given ct all atoms are injected followed by all pseudoparticles if any This is also the order in which the particles appear in trajectory output For alchemical systems the order is that of the internally constructed alchemically combined ct block 18 1 3 Force field sections Bonded and nonbonded interactions between particles are determined by the contents of the force field section of the structure file Desmond requires that each ct block except the full_system block contain a sub block named ffio_ff containing at least two sub items e an array block called fio0_sites whose attributes are summarized in Particle properties obtained from ffio sites block and e a string attribute named ffio_comb_rule the value of which specifies how Lennard Jones
97. der as the input array This behavior should be familiar to users of software packages like MATLAB Binary Threaded functions take exactly two arguments and represent an underlying function of two scalar arguments If the two arguments to the Binary Threaded function are the same length then the n th element of the return value is the underlying binary function applied to the n th elements of each of the two arguments For example a b is just the element wise sum of a and b If a binary threaded function has an argument of length 1 then that argument is paired with each of the elements of the other argument For example the return value of ax 5 is the array whose n th element is 5 times the n th element of a The behavior of Binary Threaded functions is similar but not identical to MATLAB s treatment of addition Special Forms evaluate some or all their arguments in a non standard manner The output statement print is an example The documentation for these functions explain their argument evaluation rules 11 3 5 Functions Below is a list of the available functions with brief descriptions of their behavior Full descriptions of the functions are available in Enhanced sampling function reference Table 11 1 Brief description of enhanced sampling functions x multiplication addition subtraction division a raise to integer power acos arccosine angle cosine of angle for 2 vect
98. description rO FLOAT equilibrium separation LENGTH fc FLOAT force constant ENERGY LENGTH po INTEGER Ist particle pl INTEGER 2nd particle constrained INTEGER if nonzero constrained default 0 Stretch terms that overlap with constraints should have the constrained field set to 1 Applications that evaluate constraint terms need not evaluate st ret ch_harm records that are marked as constrained These terms are in the st ret ch Hamiltonian category 7 2 2 Angle terms The angle vibration between three atoms i j k is evaluated as Va Bigk FelOijx 90 where f is the angle force constant in Energy Radians and o is the equilibrium angle in radians Beware the explicit use of the 0 angle will introduce discontinuities in the potential at 0 7 Terms in angle_harm are evaluated using this potential 7 2 Bonded pair and excluded interactions 51 Desmond Users Guide Release 3 4 0 0 7 Table 7 3 Schema for the angle_harm table name type description thetaO FLOAT equilibrium angle DEGREES fc FLOAT force constant ENERGY RADIAN po INTEGER Ist particle pl INTEGER 2nd particle p2 INTEGER 3rd particle constrained INTEGER constrained if nonzero default 0 The p0 particle forms the vertex Angle terms that overlap with constraints should have the constrained field set to 1 Applications that evaluate constraint terms
99. dification provided tha be distributed by the LICENS any 4 Acknowledgement and Citation LICENSEE agrees to acknowledge the use of the SOFTWARE in any reports or publications of results obtained with the SOFTWARE as follows Desmond Molecular Dynamics System version X Y D E Shaw Research New Your NY 2008M Where X and Y are to be replaced with the major and minor release number of the version used in based on resul code not developed by DESR EF such LICENS is paper K J Bowers Chow B S H Xu R Os the published research ts obtained with any Software Modification or any complementary then those variants must be acknowledged as also requested to include a citation to the following Dror If the published research is Min Ei Eastwood B A Gregersen 156 Chapter 20 Licenses and Third Party Software Desmond Users Guide Release 3 4 0 0 7 J L Klepeis I Kolossvary M A Moraes F D Sacerdoti J K Salmon Y Shan and D E Shaw Scalable algorithms for molecular dynamics simulations on commodity clusters Proceedings of the 2006 ACM IEEE Conference on Supercomputing SC06 Tampa FL 11 to 17 November 2006 ACM Press New York 2006 5 Disclaimer of Warranties and Liabilities LICENSEE acknowledges that the SOFTWARE is a
100. e for debugging mdsim vrun minimize remd Each application reads a particle system and a force field from a structure file located at the path p the details of which can be found in Preparing a structure file The structure file defines the global cell dimensions initial particle properties and the specific parameters of the force field Many Desmond objects share the following configuration idiom 15 Desmond Users Guide Release 3 4 0 0 7 object first tf interval t This describes the pattern of activity of the object acting only at specific times the first time at tf and thereafter periodically with period Setting t O causes the object to act at every opportunity after tp Note The application might modify 7 and t slightly from their configuration values to make them a multiple of the current timestep Setting tf to inf meaning infinity see Appendix Appendix Units declares that the activity never occurs but beware some plugins use the Boolean parameter write_last_step that when set causes output to occur at the end of the simulation regardless 4 2 Invoking Desmond Desmond applications are invoked from the command line by the desmond executable Use the include to specify the configuration file For example to invoke desmond with the configuration file equil cfg desmond includ quil cfg As indicated above the configuration s
101. e pi 3 14159 file myDoc2 txt matrix 1 0 0 0 1 0 1 1 1 options verbose no Nsteps 50 124 Chapter 15 Appendix Configuration syntax CHAPTER SIXTEEN APPENDIX CLONE RADIUS RESTRICTIONS This appendix provides the full set of restrictions on the size of the clone radius for those who need more than the practical guidelines given in The Global Cell The clone radius must be chosen large enough to ensure that a process can access all the particles it needs to compute force interactions There are however also practical limits on the size of the clone radius This Appendix collects all the restrictions placed on the clone radius For correct pairlist reconstruction Desmond requires 2Rolone 2 Riazy Rout A recall that Reut is a parameter in force nonbonded and A is global_cell margin This is normally how the clone radius is chosen it is set to half of the lazy radius plus a small fudge factor of about 107 to allow for roundoff error To correctly compute bonded interactions and constraints Relone Should be large enough that every such group of bonded or constrained particles fit within some sphere of radius Relone When a violation of this condition would prevent correct computation Desmond halts with an error For practical values of the cutoff radius Reut lt Riazy lt 2 Relone Relone Should be large enough to guarantee that each process has all the particles it requires for bonded force an
102. e is described in Migration But if particles aren t reassigned to new processes every time positions are updated then inevitably between migration events some particles will approach each other and drift within the cutoff radius Then the near interactions between the pair will have to be calculated How often this happens depends on the size of the cutoff radius and how volatile the simulation is the faster particles move the more often pairs of particles will end up in separate home boxes For efficiency Desmond maintains a list the pairlist of particle pairs that might need to be used to evaluate the effects of nonbonded near interactions The pairlist must contain particle pairs that are now outside the cutoff radius but might approach each other closely enough to interact in upcoming timesteps before the next migration Instead of the cutoff radius therefore the pairlist contains particle pairs separated by less than the lazy radius The lazy radius sets the maximum distance of all pairs of particles included in the pairlist at the time of its assembly the most recent migration The lazy radius is determined implicitly from the margin parameter A by Riazy Reut A If no particle has moved a distance more than A 2 since the last update the pairlist still contains all pairs of particles within Reut of each other In typical simulations it is highly unlikely that particles move faster than 50A ps by a probabilistic argumen
103. e application itself but you can include your own plugins in the application by implementing them in an extension a shared library so file which is dynamically linked into an application at runtime All plugins for Desmond must be organized into extensions You can create an extension with nothing more than GNU make To create an extension e Put the root of the Desmond tree containing the plugins subdirectory into the include path and add include lt Desmond Desmond hxx gt to the top of the extension s header file e Compile and link the plugin as a shared library without linking against any Desmond libraries Be sure to compile and link with fP IC required in Linux when loading shared libraries e Other compiler flags and preprocessing directives may have to be set in accordance with the particulars of the Desmond installation This may require recording the flags passed to Desmond during installation unfortu nately e Extensions are loaded into Desmond with RTLD_GLOBAL so place all classes defined by the extension into either an anonymous namespace or a namespace unique to your development environment e If you wish to checkpoint your simulation all API subclasses must be serializable These classes need to follow conventions layed out in base desmond_src util desurrection 12 1 1 Plugin interface Desmond provides a number of APIs which can be extended to provide additional functionality following the abstract factory
104. e code 11 1 2 Enhanced sampling functionality The enhanced sampling plugin is capable of performing umbrella sampling for potentials that can be expressed as functions of the coordinates of a subset of particles expressed as VMD selections To support complex potentials a simple interpreter for symbolic expressions has been developed The interpreter allows the user to specify the potential using a set of primitive operations such as norm and arithmetic operators which will be transformed into a Desmond configuration file The advantage of symbolic expressions is that the user needs only to specify the potential and the force associated with the potential will be calculated automatically The expressions may also include more complex primitives such as RMSD computation that specialize the expressions to handle common chemistry potentials It is expected that the number of available chemistry specific primitives will grow as awkward or frequently used constructs are identified Metadynamics is supported through the same interpreter as umbrella sampling and the collective coordinates needed for metadynamics are specified using the same symbolic expressions The metadynamics coordinates may also be arbitrary functions of the particle positions so long as they are expressible using the expression primitives Support for metadynamics is provided through the meta keyword in the symbolic expressions and umbrella sampling may be used in conjunction wit
105. e format using current simulation coordinates It preserves non coordinate information from the structure file Configuration information is given in app plugin list T key key type maeff_output first ty interval ti name p write_last_step bi periodicfix bp 32 Chapter 4 Running Desmond Desmond Users Guide Release 3 4 0 0 7 Table 4 15 Configuration for maeff_output name description first First time for this action Time interval Time between actions Time name The output file name String write_last_step Whether to write a structure file at the last step Boolean periodicfix Whether to wrap atom positions across periodic boundaries to minimize bond lengths Boolean 4 6 8 posre_schedule The posre_schedule plugin scales the strength of position restraints according to a time schedule It is useful for slowly turning off position restraints during a simulation The following Example shows the configuration app plugin list css Key ss J key type posre_schedul schedule time ti to tn J value Si So Sn Table 4 16 Configuration for posreschedule name description time Times at which scale factors are specified List of Times value must be same Scale factor to apply to position restraints Required length as time list List of Scalars The scale factor S used at time t i
106. e interactions refer back to records in the particle table References to particles should follow a naming convention of p0 pl p2 for each particle referenced 127 Desmond Users Guide Release 3 4 0 0 7 17 1 1 Particles The particle table associates a unique id to all particles in the structure The ids of the particles must all be contiguous starting at zero The ordering of the particles in a DMS file for the purpose of e g writing coordinate data is given by the order of their ids The minimal schema for the particle table is given in Schema for the particle table Table 17 2 Schema for the particle table name type description 1d INTEGER unique particle identifier anum INTEGER atomic number xX FLOAT x coordinate in LENGTH y FLOAT y coordinate in LENGTH Z FLOAT z coordinate in LENGTH 17 1 2 Bonds Table 17 3 Schema for the bond table name type description po INTEGER Ist particle id pl INTEGER 2nd particle id order FLOAT bond order The bond table specifies the chemical topology of the system Here the topology is understood to be independent of the forcefield that describes the interactions between particles Whether a water molecule is described by a set of stretch and angle terms or by a single constraint term one would still expect to find entries in the bond table corresponding to the two oxygen hydrogen bonds Bonds may also be p
107. e objects are however pickleable The generictrajectory trajectory objects also have simplified time accessors For instance you can access and iterate from frames based on their times for example To look at all frames whose times are between 20 5 and 30 5 from generictrajectory import T Trajectory path to somewhere dtr for frame in T at_time_ge 20 5 if frame time gt 30 5 break process frame The iterators can be accessed via the at_time_near at_time_lt at_time_le at_time_gt at_time_ge methods 13 4 3 molfile module The mol file module is a Python interface to the set of file I O plugins that are included with the program VMD de veloped at the University of Illinois The Python interface provides methods for creating loading and saving molecular structures and coordinates to all the file formats supported by VMD Below is a synopsis of how to perform common tasks using molfile import molfile 116 Chapter 13 Trajectory Format and Analysis Desmond Users Guide Release 3 4 0 0 7 Reading a structure file reader molfile mae read path to foo mae Iterating through the frames in a file for frame in molfile dtr read path to foo dtr frames function frame pos frame vel frame time frame box Random access to frames only dtr files support this currently 27 molfile dtr read path to foo dtr frame 27 O based index Convert an mae file to a pdb file
108. e vectors S and is a standard Wiener process and op V 2W kp Thro With the NoseHoover thermostat the barostat steps add two variables r p s 1 variables as a Martyna Tobias Klein piston t v for each 7 and evolve them and the 7 A y r W i msi W amp uiw pi 1 1 Ng A n pi W a re2s6 PBODBO y D E mi w i mi b 2 b b bj b p 5 mn W CP vjva wa i j X2 Ja j dd J 11 wi_4 Ci Vis Wiy1 i vi t a an v where each of the components of the vectors S and S is a standard Wiener process and op vV 2W kp Thro With the Antithetic thermostat the barostat steps are as in the unthermostatted case albeit the 7 variables are governed by the discrete antithetic dynamics described in Anti_NVT Antithetic constant volume and temperature 9 6 Dynamical systems 89 Desmond Users Guide Release 3 4 0 0 7 Table 9 14 Configuration for Multigrator name description nve type the type of NVE step none Verlet PLS thermostat type the type of thermostat step Langevin NoseHoover Antithetic Mixed thermostat timesteps Number of innermost time steps per full thermostat step Integer gt 0 a multiple of the outer RESPA timesteps barostat type the type of barostat step MTK barostat thermostat type thermostat type of the barostat step Langevin NoseHoover Antithetic barostat timesteps Number of innermost time steps per full b
109. each replica The only restriction on the replicas themselves is that they must all have the same number of particles Thus remd can be used for the usual temperature exchange method as well as exchanges between systems with different Hamiltonian parameters remd runs as a single parallel application just like mdsim and vrun producing a single checkpoint file if check pointing is enabled Each replica runs as a normal simulation with swaps of coordinates taking place as specified by the user through the configuration When an exchange is attempted between two replicas the usual Metropolis crite rion is applied to determine if the exchange should be accepted or accepted according to the following prescription with Q 61U11 b2U22 b1U12 b2U21 81 Pi B2P2 Vi Vo 4 1 where randy is a random variate on 0 1 U is the potential energy of replica i in the Hamiltonian of replica j P is the reference pressure of replica i V is instantaneous volume of replica 7 and 6 is the inverse reference temperature of replica i If Q gt 0 accept the exchange or if Q lt 20 reject it otherwise accept the exchange if randy lt exp Q An example remd configuration is shown in following Example all parameters are required The parameters are summarized in Configuration for remd remd title w last_time t checkpt 1 plugin first tf interval ti seed s xchange_type neighbors random c
110. each slab totaled over categories and kinetic Some force components especially far_terms are expensive to compute and vary slowly with time One can im prove the efficiency of the pressure profile calculation by instantiating the pprofile plugin twice with one instance evaluating the non far_terms components relatively frequently and the other instance evaluating the far_terms components relatively infrequently For example app plugin list slow fast slow type pprofile first 0 interval 10 eval_interval 0 2 nslabs 8 name pp slow dtr include far_terms 34 Chapter 4 Running Desmond Desmond Users Guide Release 3 4 0 0 7 fast fast type pprofile first 0 interval 10 eval_interval 0 01 nslabs 8 name pp fast dtr exclude far_terms 4 6 10 randomize_velocities The randomize_velocities plugin periodically thermalizes velocities Configuration is shown in app plugin list key key type randomize_velocities first tf interval ti seed Ss temperature T You can use this plugin to perform initial velocity randomization by setting the value of first to zero interval to infinity and temperature to the desired temperature The plugin can also serve as a rough implementation of an Andersen thermostat Table 4 18 Configuration for randomize_velocities name description first First time for this actio
111. ecause these applications are expected to run for long periods of time during which hardware might fail they can be set to produce a checkpoint file periodically from which you can restart 4 2 1 Restoring from a checkpoint You can configure the mdsim or remd applications to create a checkpoint file at regular intervals as it runs When you wish desmond to start from a checkpoint file created during an earlier run use the restore flag to specify the file name For example to restore from a checkpoint 4 2 Invoking Desmond 17 Desmond Users Guide Release 3 4 0 0 7 desmond tpp 4 restore checkpoint_file Note To avoid an application error set the t pp and other thread specific flags the same way it was set for the original simulation desmond must initialize the parallel environment before it can read the checkpoint file You need not specify other configuration options they ve been saved When restoring from a checkpoint file only certain options can be changed from the configuration of the original simulation Last _t ime see mdsim and remd checkpt interval see Checkpointing and certain plugin options for example the name and interval for eneseqandtrajectory 4 2 2 Using plugins Desmond applications use certain plugins for various diagnotics and interventions Plugins can be implemented as part of an application called built in plugins or in external files called extensions Desmond loc
112. ed that the following conditions are met Redistributions of source code must retain the above copyright notice this list of conditions and the following disclaimer Redistributions in binary form must reproduce the above copyright notice this list of conditions and the following disclaimer in the documentation and or other materials provided with the distribution Neither the name of the University of Cambridge nor the name of Google Inc nor the names of their contributors may be used to endorse or promote products derived from this software without specific prior written permission THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS AS IS AND ANY EXPRESS OR IMPLIED WARRANTIES INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF ERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSI ARE DISCLAIMED IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT INDIRECT INCIDENTAL SPECIAL EXEMPLARY OR CONSEQUENTIAL DAMAGES INCLUDING BUT NO LIMITED TO PROCUREME OF SUBSTITUTE GOODS OR SERVICES LOSS OF USE DATA OR PROFITS OR BUSINESS INTERRUPTION HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY WHETHER IN CONTRACT STRICT LIABILITY OR TORT INCLUDING NEGLIGENCE OR OTHERWISE ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE El
113. eference temperature Temperature gt 0 9 6 6 MTK_NPT Martyna Tobias Klein constant pressure and temperature The MTK_NPT dynamical system Mar 1994 is configured as shown in integrator MTK_NPT barostat tau Tp T_ref Tb thermostat 80 Chapter 9 Dynamics Desmond Users Guide Release 3 4 0 0 7 mts m tau T is T thermostat mts m tau Ti s Tn The Martyna Tobias Klein dynamical system is a combination of Piston_NPH see Piston NPH constant pressure and enthalpy and NH_NVT see NH_NVT Nos Hoover constant volume and temperature dynamics There is also an additional Nos Hoover chain with n additional pairs of variables C v that govern the barostat degrees of freedom To include this chain in sums or products over chains treat the index of the sum or product as ranging over the numbers 1 k for the particle Nos Hoover chains and the letter b The ODE for this system is Ti Pi mi A n ri W i nisi W G viu Vz U r B s 1 1 N A M W piv JO haa ec p B s Po B s B s mA gt am mvt ju ith gt j yd fay we Y Blm Cf 3 3 ilx i j Soni W C viva w i 5 PN pg j if WLP fwhy Cl virla fola v 1 w a ci where C kpTpd and C i gt 1 kBTp where d is the number of independent variables in the barostat according to its isotropy type and wi
114. emperature cv dist p 1 p 0 collective variable is interatomic distance meta 0 array h_0 exp meta 0 array 0 0 cv kT1 w cv 11 5 4 Metadynamics with a wall This example demonstrates the use of a wall to prevent metadynamics from driving the collective coordinates too far The form of this wall is Await 1 exp 2 Wwall where hwan is the wall height xo is the location of the wall c is the collective variable and wwan is the width of the wall The wall potential is added as an umbrella potential to the enhanced sampling symbolic expression declare_meta dimension 1 cutoff 9 first 0 0 interval 0 200 name kerseq initial p atomsel index 0 1 cv dist p 11 pl0 collective variable x0 14 wall location 11 5 Examples 107 Desmond Users Guide Release 3 4 0 0 7 w_wall 0 2 wall width h_wall 1000 wall height wall h_wall 1 exp x0 cv w_wall wall meta 0 array 0 2 0 1 cv 108 Chapter 11 Enhanced Sampling and Umbrella Sampling CHAPTER TWELVE EXTENDING DESMOND This chapter provides a sketch for implementing extensions for Desmond Full technical specifications are difficult to accomplish or keep current in a document removed from the source files Hence this chapter can only provide an outline and some pointers for further information 12 1 Implementation Desmond s built in plugins are compiled with th
115. energy properly a function of both r and the 3 x 3 matrix B a b where a b and are the lattice vectors of the cell Usually this dependence on B is suppressed unless variations in the cell shape need to be considered 9 1 2 Chemical systems In addition to the energy of the particles a number of other macroscopic properties of the system are of interest particularly pressure and temperature These quantities are only properly defined in reference to very large systems with ergodic dynamics aver aged over statistically significant lengths of time However instantaneous microscopic versions of these quantities can be defined The instantaneous temperature T of a group of particles is given by 1 1 kpT 711 2m aha yy DI 2me where kg is the Boltzmann constant and Ny counts the number of degrees of freedom of the particles for N free particles N 3N The instantaneous pressure is given by P Tr P 3 the average of the main diagonal components of the 3 x 3 tensor P r p B B p p mi Vr U r BY VgU r pa t Variations of the Newton equations are often made through additional ordinary or stochastic variables coupled dynam ically to the positions and momenta or via feedback control interventions which adjust the positions and momenta These variations are typically designed to ensure certain statistical properties of the macroscopic quantities 9 2 Integrator Simulation dyna
116. ength For the purpose of derivative computation the derivative of a static variable is always zero even if the value stored had a nonzero derivative As an example of two uses of static variables the following potential restrains a particle to its initial location and prints the displacement vector of the particle on this time step static x0 3 x_last 3 k 10 p atomsel index 10 x pos p 0 store x_last x print x_diff min_image x x_last if time then k x norm2 min_image x x0 else store x0 x 102 Chapter 11 Enhanced Sampling and Umbrella Sampling Desmond Users Guide Release 3 4 0 0 7 0 y Note that the printed difference will not make sense for the initial step because static variables are initialized with Zeros 11 3 4 Function classes There are four classes of functions which differ in the way they evaluate their arguments The classes are Normal Threaded Binary Threaded and Special Forms Unless otherwise noted arguments are evaluated in left to right order Normal functions evaluate all their arguments before the function body is entered After the arguments are evaluated the function executes with the value of the arguments Threaded functions take exactly one argument and compute their return value element wise over the elements of their argument For example if cos is applied to an array of angles the result is an array of cosine values in the same or
117. entary probability theory that the c d f is a mapping from energies in 0 00 to probabilities in 0 1 The mapping gives the probability that a random variable sampled from T will be less than or equal to a given energy The quantile function is the inverse map from probabilities to energies The effect of the antithetic thermostat is to change the total energy Ej of each energy group and its associated bath every full timestep A according to E F 1 F E This is accomplished by uniformly scaling the particle momenta p within each thermostat at each full timestep Ay F 1 F E a i E En s wherej x i This system does not preserve a scalar quantity In the absence of the NVE dynamics the discrete antithetic dynamics would preserve the phase space density 1 12 11 Q gt gt Q a p kpT 2mi 0 j ilx i j The combined dynamics however does not in general preserve this density The resulting phase space density is not easily expressed in the general case Only when T Tk T does this system preserve the phase space density Q exp Ho p keT M 76 Chapter 9 Dynamics Desmond Users Guide Release 3 4 0 0 7 Thus the trajectories of this system are expected to produce samples from the canonical ensemble with temperature T Note that the preserved density is independent of the bath size Np Increasing the bath size merely serves to decrease the magnitude of the mome
118. ere continuity of energy and derivatives dictate that a s 3s 2b c s b 2s In the above s and b correspond to sigma and beta in the stretch_fbhw schema Table 7 11 Schema for the stretch_fbhw table name type description lower FLOAT lower bound for flat bottomed region in LENGTH upper FLOAT upper bound for flat bottomed region in LENGTH sigma FLOAT width of harmonic region in LENGTH for r gt upper beta FLOAT slope of linear region in ENERGY LENGTH fc FLOAT overall proportionality constant in ENERGY LENGTH group INTEGER tag for first group with specified parameters group2 INTEGER tag for second group with specified parameters Table 7 12 Schema for the stretch_fbhw_term ta ble name type description po INTEGER particle id group INTEGER group for given particle The functional form of the flat bottomed harmonic angle term is V d where O9 0 where o lt o d lt 0 where lt 0 bo lt o 9 07 0 where o lt 0 0o and theta0 is in radians Table 7 13 Schema for the angle_fbhw table name type description fc FLOAT force constant in ENERGY RADIANS theta0 FLOAT equilibrium angle in DEGREES sigma FLOAT half width of flat bottomed region in DEGREES po INTEGER first particle pl INTEGER second particle p2 INTEGER third parti
119. ere particle is in the j th thermostat temperature T The Wiener distribution is seeded by s Although this SDE does not have a conserved scalar it does have an invariant phase space density given by Q flr p Pras where f satisfies the PDE 0 3 A Vrf VaU r Vif va pif 7 vas r If 7 Tk T then f exp Ho r p k8T 9 6 Dynamical systems 77 Desmond Users Guide Release 3 4 0 0 7 Thus the trajectories of this system are expected to produce samples from the canonical ensemble with temperature T In Desmond the net energy or heat subtracted by the stochastic portions of the SDE are accounted for in the extended variable energy term which results in a conserved energy useful for diagnostic purposes Table 9 7 Configuration for L_NVT name description thermostat tau The decay time inverse damping constant of the particle momenta Time gt 0 thermostat seed The random number seed for normally distributed random variables Integer 9 6 5 Piston_NPH constant pressure and enthalpy The Piston_NPH dynamical system is configured as shown in integrator Piston_NPH barostat tau Tp T_ref Tp optional This is the simplest dynamical system that changes the cell according to a conservative dynamics More complex systems that change the cell have many similarities with Piston_NPH and share its definitions Usually energy is the con
120. erendsen constant volume and temperature simulation the particle velocities are rescaled at each full timestep At to bring the instantaneous temperature T closer to the target temperature T if T gt Tj the particle velocities are scaled down if the T lt T the particle velocities are scaled up Velocities are rescaled gradually according to a linear rate given by 7 A AT T Tj To elaborate scaling the particle velocities by s scales the kinetic energy and instantaneous temperature by 5 thus ATx 5 DVI serves to determine sj However such a procedure tends to be unstable unless the center of mass motion of the entire system is simultaneously removed With the mass and velocity of each thermostat defined by M 5 mi ilx i j and ilx j j The velocity of the system after rescaling is gt 1 5 Vis 97 1007 where M X gt j Mj The new particle momenta are given by new Mo 8 bi mV s where s is determined by solving the following nonlinear equation 1 Es A 1 AK sK s M V V s 5 MallV s II es Storm K Table 9 11 Configuration for Ber_NVT name description tau Relaxation time Time gt 0 min_velocity_scaling Minimum factor for scaling particle velocities in one timestep 0 lt Real lt 1 max_velocity_scaling Maximum factor for scaling particle velocities in one timestep 1 lt Real thermostat Description of the thermostat for the particles see N
121. ers indicate an angle involving at least one dummy atom 0 indicates that this potential should disappear in the corresponding ct i_fepio_tj Angle potential number in perturbed_ct Negative angle numbers indicate an angle involving at least one dummy atom 0 indicates that this potential should disappear in the corresponding ct i_fepio_ai The first atom in the angle in original_ct Negative atom number can appear here by the same convention as in atommaps i_fepio_aj The second atom in the angle in orignal_ct Negative atom number can appear here by the same convention as in atommaps i_fepio_ak The third atom in the angle in orignal_ct Negative atom number can appear here by the same convention as in atommaps fepio_dihedralmaps This indexed block maps the dihedral angle potentials from original_ct onto perturbed_ct Table 18 9 fepio_dihedralmaps properties property name description 1 fepio_ti Dihedral potential number in original_ct Negative dihedral numbers indicate a dihedral involving at least one dummy atom 0 indicates that this potential should disappear in the corresponding ct 1 fepio_tj Dihedral potential number in perturbed_ct Negative dihedral numbers indicate a dihedral involving at least one dummy atom 0 indicates that this potential should disappear in the corresponding ct i_fepio_ai The first atom in the dihedral in original_ct Negative atom number can appear here by the sa
122. erspersed with momentum updates Each position update is identical and takes the form of Equation 9 2 Each momentum update takes the form of Equation 9 1 using the weighted combination Filt E t oiinf t ofisf l t where the o are both 1 for the first step and thereafter o 1 every 1 steps and 0 otherwise If there are no outer step dynamics and i kip then the full integration step is equivalent to the concatenation of k the integration steps one obtains with 7 iy We illustrate a in if io 2 4 8 schedule in the RESPA schematic example Multiple timestepping is effectively disabled by setting in if 1 1 which makes the force and extended dynamics purely a function of current position f t r t and not the phase of time 9 3 RESPA 71 Desmond Users Guide Release 3 4 0 0 7 f Los Sol odo 6 6 0 1 Figure 9 1 A schematic representation of a in if io 2 4 8 RESPA schedule Stacks of circles represent momentum updates with bonded near and far force components Squares represent position updates Table 9 2 Configuration for respa name description near_timesteps The number of position updates per nonbonded near force calculation Integer gt 0 divides far_timesteps far_timesteps The number of position updates per nonbonded far force calculation Integer gt 0 divides the outer_timesteps outer_timesteps The number of position updates per application
123. et x Redistributions of source code must retain the above copyright notice this list of conditions and the following disclaimer x Redistributions in binary form must reproduce the abov copyright notice this list of conditions and the following 20 2 Licensed Companion Software 159 Desmond Users Guide Release 3 4 0 0 7 disclaimer in the documentation and or other mater with the distribution Neither the name of Google Inc nor the names of i contributors may be used to endorse or promote pro from this software without specific prior written HIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND C AS IS AND ANY EXPRESS OR IMPLI 7j ials provided ts ducts derived permission ONTRIBUTORS D WARRANTIES INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED IN NO EVENT SHALL HE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT INDIREC SPECIAL EXEMPLARY OR CONSEQUENTIAL DAMAGES INCLUDING LIMITED TO PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES INCIDENTAL BUT NOT OSS OF USE ED AND ON ANY DATA OR PROFITS OR BUSINESS INTERRUPTION HOWEVER CAUS HEORY OF LIABILITY WHETHER IN CONTRACT STRICT LIABILI INC
124. f Equilibrium temperature see Piston_NPH and L_NVT Optional defaults to the global reference temperature Temperature gt 0 barostat thermostat Description of the thermostat of the barostat see L_NVT Langevin parameters thermostat Description of the thermostat for the particles see L_NV T Langevin parameters 9 6 8 Ber_NVT Berendsen constant volume and temperature The Ber_NVT dynamical system Ber 1984 is configured as shown in integrator Ber_NVT tau min_velocity_scaling Smin max_velocity_scaling Smax Berendsen constant volume and temperature simulations do not sample microstates according to their probability distribution in a canonical ensemble Instead this dynamics keeps the kinetic energy of the system close to the average kinetic energy in the corresponding canonical ensemble by means of feedback control It can be used to equilibrate a system in short simulations It is recommended that Berendsen integrators be run with the net center of mass motion periodically removed from the system to prevent certain long term degenerate behaviors The instantaneous temperature T of the atoms governed by thermostat j with reference temperature T is related to their kinetic energy by gt 1 Kj Y IIpil Qmi NjkaT ix s 9 6 Dynamical systems 83 Desmond Users Guide Release 3 4 0 0 7 where N is the number of degrees of freedom of thermostat j In a B
125. f several of Desmond s parallelization mechanisms The global cell is divided into regular three dimensional volumes called boxes Each box is assigned to a single Desmond process which maintains the information describing each particle located within that box Note For an efficiently parallelized simulation in Desmond we recommend no more than one process one box per processor The box encompassing the volume of space in which a particle is located is called its home box The home box determines which process owns the particle that is maintains its mass charge position velocity and other associated data Interactions between particles can cross box boundaries of course communication across box boundaries can be necessary for other reasons too This means that communication must occur between processes Interactions that require communication between processes have a strong effect on how well your simulation performs in parallel how much it can take advantage of the multiple processes available to it Communication between processes is necessary to resolve two common situations e A particle near the face of a box is bonded with a particle in a neighboring box or close enough to it that the electrostatic or van der Waals forces between them are computed explicitly that is within the cutoff radius see Space e A particle that was not originally inside the cutoff radius drifts inside it from one timestep to the next
126. fg i Cp 4 4 Configuring Desmond applications 21 Desmond Users Guide Release 3 4 0 0 7 Table 4 4 Configuration for remd name description title A short string to include in various output files Optional by default no title string last_time Time at which to stop the simulation in picoseconds relative to the reference time given as part of the global cell configuration see Configuration time checkpt Checkpoint configuration See Checkpointing configuration plugin See Using plugins configuration Est Time of first exchange attempt Time interval Time between exchange attempts Time type Either exchanges only between neighboring replicas or exchanges between any pair of replicas neighbors random seed random number seed for the Metropolis criterion Integer cfg configuration overrides for each replica List of configurations Exchanges are attempted starting at chemical time given by first and at intervals of interval thereafter If type is neighbors then on each exchange attempt all replicas will attempt an exchange with either of their neighbors in a linear order with 50and accept based on the Metropolis criterion above If type is random then only two out of all replicas will attempt an exchange but those two replicas could be any of the replicas in the ensemble Exchanges are implemented by swapping the positions of a pair of replica
127. field data contained in the frames of a frameset Desmond can write either its floating point positition data in a bitwise precise internal form or a simpler to access floating point form In its simplest form Python framesets provide a frame iterator and numpy array access to data fields import framesettools fs framesettools FrameSet myframeset dtr print myframeset has len fs frames from time fs times 0 to fs times 1 assumes a normal WRAPPED_V_2 Desmond trajectory for frame in fs x frame POSITION 0 3 print frame CHEMICALTIME atom 0 has position x Users can also write w overwrite w or extend a trajectories By default framesets opened in write mode will fail if the file already exists Use w if you wish to rewrite an existing frameset Here is a sample program that will randomize positions import framesettools import random fs framesettools FrameSet foobar dtr out framesettools FrameSet output dtr w for frame in fs This iterates over all the frames pos frame POSITION This is a 1 D 3 natoms numpy array delta random gauss 0 1 for i in range len pos pos delta out push_back frame frame CHEMICALTIME An example that writes out all the ENERGY fields import framesettools import random fs framesettools FrameSet foobar dtr out framesettools FrameSe
128. first tf interval ti name filename weights The free energy of a thermodynamic system with Hamiltonian H is related to the partition function Z of the corresponding ensemble by F kgT In Zp where kg is the Boltzmann constant and T is the temperature and Zy is the partition function for the Hamiltonian H The free energy is not an average of some quantity over the phase space therefore it can not be computed from molecular dynamic simulations or other importance sampling techniques Fortunately what matters in problems of chemistry and biology is the relative free energy the difference between two systems acting through different Hamil tonians This difference in free energy can be expressed as an ensemble average and is thus amenable to computation by importance sampling Consider two systems with different Hamiltonians Ho and H In the canonical ensemble at temperature T the 93 Desmond Users Guide Release 3 4 0 0 7 free energy difference between the two systems 1s F Fy Fo kgT ln Zo Z1 kaT In Ze d7 2 kaT ln J Zg eP efo E TT ar kT lh AED 0 where d37 is the volume elements of the position of particle i This equation suggests that at least in theory we can compute AF by sampling r according to the canonical dis tribution e 0 and computing the average of eP Ho H1 2 In practice we use better estimators such as the Bennett acceptance ratio
129. front of an expression Importantly the subtraction operator does not perform a minimum image computation See Periodic Images for more information Array subscripts are indicated by using the syntax ali where both the array and the index may be arbitrary expressions Array subscripts have higher precedence than the binary arithmetic operators Expressions may indicate conditionals with the notation if condition then positive branch else nonpositive branch Note that if returns a value and may be used in expressions The condition must be a single number and the positive branch is used if that number is greater than zero An example is interaction if time 10 then k x 2 if time gt 10 use harmonic potential else 0 otherwise use zero potential The unneeded branch is not executed The only looping construct in the m expression language is the series expression which sums its body over a set of iterators As an example the following series computes the sum of all harmonic pairwise interactions between sets of particles a and b s series i 0 length a j 0 length b k dist alil b j 2 Each iterator is specified as iter_name lower_bound upper_bound and the iteration is carried out for all integers i where lower_bound lt i lt upper_bound Expression blocks and scoping are available Blocks are indicated by wrapping a sequence of statements in braces and blocks may appear anywhere within expressio
130. g the initial steps of the simulation and becomes superfluous later When simulating polarizable force fields with damped Drude dynamics the Brownian dynamics integrators use a minimizer to reposition the Drude particles after each timestep This minimization is carried out using the same quasi Newton minimizer that is applied at initialization This improves stability and accuracy as the algorithm employed in damped Drude dynamics is based upon the smooth evolution of the exact locations of the minima As per the corresponding Langevin integrators the Brownian dynamics integrators track the net heat transferred from the stochastic processes in their extended variable energies which creates a useful conserved diagnostic quantity The velocity clipping process which removes kinetic energy from the system is not accounted for and will cause the diagnostic quantity and extended variable energy to increase when clipping takes place Table 9 13 Configuration for brownian name description delta_max maximum displacement of any particle position per step Length gt 0 thermostat seed random seed for normally distributed random variables of the particles Integer barostat thermostat seed random seed for normally distributed random variables of the global cell Integer 86 Chapter 9 Dynamics Desmond Users Guide Release 3 4 0 0 7 9 6 11 The Multigrator integrator integrator Multigrator nve
131. ge 163
132. ges is established remd graph performs replica exchange by selecting an edge at random from the full set of edges remd graph gives users the option of reporting the energy differences between all pairs of configurations in the edge list of the graph The timing of the output is controlled by the and t parameters according to the usual conventions and written to the path p For each edge say T1 T2 for example a pair of values AE AE_ is reported where AF U12 U11 AE_ Uz U22 and the U are the potential energies from Equation 4 1 Table 4 5 Configuration for remd graph name description graph edges A list of edge descriptions describing the edge set of the graph List deltaE Description of the deltaE output Optional by default disabled configuration deltaE first Time of first output Time deltaE interval Time between outputs Time deltaE name Output filename to use for writing the output file filename 4 4 4 minimize minimize performs steepest descent minimization followed by LBFGS minimization Configuration parameters are shown in following example all parameters are optional the defaults should be adequate for most systems minimize migrate_interval 1 m m maxsteps Smax tol 7 stepsize l switch y sdsteps debug dt t plugin f sic So d 4 4 Configuring Desmond applications 23 Desmond Users Guide Release 3
133. gle is formed by the planes p0 p1 p2 and pl p2 p3 Terms in improper_harm are handled by 52 Chapter 7 Calculating Force and Energy Desmond Users Guide Release 3 4 0 0 7 The harmonic dihedral term given in Equation 7 2 can lead to accuracy issues if f is too small or if initial conditions are poorly chosen due to a discontinuity in the definition of the first derivative with respect to in jjx1 near Hp T Table 7 5 Schema for the improper_harm table name type description phi0 FLOAT equilibrium separation DEGREES fc FLOAT force constant ENERGY DEGREE po INTEGER Ist particle pl INTEGER 2nd particle p2 INTEGER 3rd particle p3 INTEGER 4th particle These terms are in the improper Hamiltonian category 7 2 5 CMAP torsion terms CMAP is a torsion torsion cross term that uses a tabulated energy correction It is found in more recent versions of the CHARMM forcefield The potential function is given by n 1 _ or m 1 As 4 4 Sa ZYL velt 7 Y Cam OS n l m 1 where Cnm are bi cubic interpolation coefficients derived from the supplied energy table is the dihedral angle formed by particles p0 p3 and y is the dihedral angle formed by particles p4 p7 The grid spacings are also derived from the supplied energy table Terms in torsiontorsion_cmap are handled by this potential function The cmap tables for each term can be found in cmapN
134. gma configuration parameter in the far subsection of the nonbonded subsection of the force section of the configuration file Note The internal data structure used to implement configurations is called ark and error messages referring to it most likely indicate either bad syntax or missing values In addition a configuration can include comments A comment starts with a mark and continues until the end of the line 121 Desmond Users Guide Release 3 4 0 0 7 When producing a configuration with either the include and cfg options discussed in nvoking Desmond the contents of files for the former and string arguments for the latter are concatenated and parsed as a single text with comments removed The text is parsed according to a more complex grammar CONFIG gt KEYVAL x KEYVAL gt KEYPATH VALUE KEYPATH CONFIG INCLUD KEYPATH gt key INDEX x INDEX gt number key VALUE gt CONFIG VALUE QATOM Q I El ATOM gt atom atom atom atom CLUDE gt include QATOM QATOM Resolves to either a quote delimited string using any of the standard quotation marks or a bare string a sequence of characters containing no white space or syntactic tokens Within a quote delimited string internal quotes can be escaped with a backslash as per the common convention The KEYVAL terms are interpreted in the order g
135. h rent or make the SOFTWARE or any portion thereof or outsourcing service remove or or decompile the by revers Source Code from the object code sell transfer publish display available to third parties in a service bureau use the SOFTWARE for the alter any proprietary rights notices re export the SOFTWARE of any jurisdiction the SOFTWARE D in violation of e disclose or any code RELAT ICENS shall horized access ERIALS ICENS f the foregoing ED E hereunder to or E ELAT ED MAT SR ES its affiliates and or its be an inadequate remedy and or its licensors may hreatened breach o Work Permissions ENS F the following types of derivative wor interoperates with the SOFTWARE harge and dis Shaw Research users free of c states that D approve or authorize such comprising the SOFTWARE such Software Modifica B itself code and KS Ii provided tributed only LLC and Software tion may in no case 1 and LICENS seek and obtain LICENSEE s other legal and equitable remedies relate only to any and permit LICENS to create only any complementary code that that any such code is provided to with a disclaimer that conspicuously its affiliates did not create any modification to the code Mo
136. h rigid water table respectively Each record in the AHn table gives the length of the bonds between a single parent atom and n child atoms Each record in the HOH table gives the angle between the two O H bonds and the respective bonds lengths 65 Desmond Users Guide Release 3 4 0 0 7 Table 8 1 Schema for the constraint_ahN ta bles name type description rl FLOAT A H 1 distance r2 FLOAT A H2 distance rN FLOAT A HN distance po INTEGER id of parent atom pl INTEGER id of H1 p2 INTEGER id of H2 pN INTEGER id of HN Table 8 2 Schema for the constraint_hoh rigid water table name type description theta FLOAT H O H angle in DEGREES rl FLOAT O H1 distance r2 FLOAT O H2 distance po INTEGER id of heavy atom oxygen pl INTEGER id of H1 p2 INTEGER id of H2 A constrained particle is no longer free each such particle has 3 m 2 degrees of freedom where m is the number of independent constraints involved for example a pair of particles having only one distance constraint between them has five degrees of freedom Constraints thus affect the calculation of the instantaneous temperature and pressure which depend on the number of degrees of freedom Constraints are implemented in Desmond by the M SHAKE algorithm iteratively obtaining corrections to particle positions as well as secondary corrections to momenta The implementation is
137. h metadynamics e g to provide walls to bound the collective coordinates It is important to understand that whether being used for umbrella sampling or metadynamics the action of the en hanced sampling plugin is always applied at the outer RESPA timestep For a discussion of how Desmond applies plugins see the Desmond User s Guide 99 Desmond Users Guide Release 3 4 0 0 7 11 2 Using the Enhanced Sampling Plugin 11 2 1 Workflow Enhanced sampling potentials must be specified using the imperative m expression syntax described below The user will then run the enhsamp program to transform the potential description into an s expression form suitable to use as a Desmond configuration file The enhsamp is also responsible for resolving VMD atom selections using a Maestro structure file specified in its command line arguments A typical usage of the enhanced sampling plugin is given below edit enh pot write potential file enhsamp structure file dms enh pot gt enh ark run parser mdsim include desmond config ark include enh ark launch Desmond The output of the enhsamp program is a valid Desmond configuration file and including the enhsamp output with a standard Desmond configuration file is all that is required to use the enhanced sampling plugin The structure file used with Desmond must be the same as the structure file given to the enhsamp program 11 2 2 Output format By default the only output generate
138. h other in displacement distance and or orientation It can also be used to restrain the position and orientation of a group of atoms within the molecular system with respect to the simulation box Unlike most plugins its configuration is given in the force section of the configuration note below 26 Chapter 4 Running Desmond Desmond Users Guide Release 3 4 0 0 7 force term list key ses key type BiasingForce cm_moi groups AB displace_coeff ke ky kz displacement Zo Yo Zo distance_coeff kg distance Ro orient_coeff wi w2 w3 Euler_angles 00 Po Wo use_lab_frame_for_displacement Boolean pull_displacement Uz Uy Uz pull_distance Ud pull_Euler a e de Multiple biasing potentials supplied as a list can be applied output first tf interval t name filename to t0 Other force terms The units of the parameters in the configuration are given in Table tbl BiasingForceUnits Table 4 9 Units of the parameters in biasing force quantity unit t0 picosecond ke ky kz kcal mol Zo Yo Z0 ka keal mol A Ro A W1 W2 W3 kcal mol 0o do Yo degree not radians Ur Uy Uz A picosecond Va A picosecond ae ae EL degree picosecond The biasing force in the above configuration will restrain particles in cm_moi group B with respect to particles in cm_moi gr
139. harmm27 f tip3p or f charmm27 f tip4p respectively as command line options When multiple force fields match a given residue in the structure the first forcefield takes precedence All specified force fields must have consistent van der Waals combining rules water models can be used with any force field When a bond exists between two residues both residues must be matched by exactly one of the specified force fields 46 Chapter 6 Preparing a structure file Desmond Users Guide Release 3 4 0 0 7 Table 6 1 Force fields built into Viparr Force field name Description amber03 Amber amber94 Amber amber96 Amber amber99 Amber amber99SB Amber amber99SB ILDN Amber with modifications charmm22nocmap CHARMM 22 without CMAP terms charmm22star CHARMM 22 with modifications charmm27 CHARMM 27 charmm32 CHARMM 32 charmm36_lipids CHARMM 36 lipids charmm36_nucleicacids CHARMM 36 nucleic acids oplsaa_impact_2001 OPLS AA 2001 oplsaa_impact_2005 OPLS AA 2005 spc Water model spce Water model tip3p Water model tip3p_charmm Water model tip4p Water model tip4pew Water model tipSp Water model 6 2 3 Adding constraints Like other force field terms constraint terms must be specified explicitly in this way Desmond is unlike other molecu lar dynamics applications that infer the existence of constraints based on molecular topology and configurati
140. hat is particles that move leftwards out of the global conditions cell appear to be moving in at a corresponding spot on the right hand face and vice versa particles that move out the top appear to enter at the bottom and vice versa and finally particles that move out the front appear at the back and vice versa Thus you can picture your simulation as an arbitrarily large space tiled by the global cell repeating periodically Because the global cell tiles the simulation volume it must be a shape that can tile a three dimensional space without gaps such as a parallelepiped a hexagonal prism or a truncated octahedron The global cell also has specified dimensions It must be large enough that the molecule of interest doesn t interact with its counterparts its periodic images in other repetitions of the global cell When you run a simulation in parallel Desmond apportions the work among processes by breaking the global cell into smaller boxes Therefore how you configure the global cell can have a significant effect on how efficiently your simulation runs in parallel Details of these parallelization parameters and related ones are discussed in Configuration 3 6 Time The simulation begins at a specified reference time and advances by timesteps The time at which the simulation begins Ordinarily a simulation begins at time 0 0 but it need not For example if you wish to use the output of one simulation as the input for the nex
141. he last specified name for the structure file overrides any previous ones 16 Chapter 4 Running Desmond Desmond Users Guide Release 3 4 0 0 7 The t pp command line option sets the number of threads per process If your application is to run on a processor with multiple cores you may benefit by setting this value to other than its default of one Otherwise the command line can omit it The cpc command line option sets the number of cores per physical chip and as a side effect ties Desmond threads to processor cores If cpc N where N gt 1 is used master and worker threads are bound to processor cores If spin 1or spin 2 is used a faster but more processor intensive thread idle strategy using spin locks is employed When 1 foreground threads will spin and background threads will sleep when 2 all worker threads will spin Note If you run more than one Desmond job on a multiprocessor node make sure that cpc is set to 0 otherwise Desmond processes in the different jobs will use the same core resulting in significant performance degradation Note When running on an interactively used workstation and with more than one Desmond thread it is better to set spin 0 For example to start a Desmond application with four threads per process desmond tpp 4 includ xample cfg cfg boot file input dms Note Under most circumstances it s best to run desmond with one thread per process and one pr
142. he mdsim application on the other hand ignores the constrained bond and angle terms and prints a message at startup indicating how many terms have been ignored 6 2 Preparing a Desmond DMS file 47 Desmond Users Guide Release 3 4 0 0 7 6 2 4 Running the build_constraints program To run build_constraints build _constraints options input dms output dms The options are e k Leave constrained bonds and angle terms unmodified rather than setting their const rained column to 1 e x C Don t build any constraints of type C 48 Chapter 6 Preparing a structure file CHAPTER SEVEN CALCULATING FORCE AND ENERGY This chapter provides a high level overview of configuring force fields then discusses the computations involved in and how to configure the various interactions It also describes additional off atom interaction sites 7 1 Configuring force fields Force fields are configured as shown in force bonded exclude wa optional include Share optional virtual exclude sau optional include ake optional constraint exclude Sears optional include asa optional polar exclude optional include optional nonbonded vdW and es term force plugins ignore_com_dofs b Many molecular force fields approximate the total potential energy of a chemical system as a sum of the form U Ubonded Uvaw
143. he unit cell atom positions and atom velocities 13 1 Structure of frameset directories Framesets are stored in standard file system directories At the top level of the directory are the timekeys file the metadata file a clickme file and the not hashed directory which holds the ddparams file The frame data is held in frame files of the form frame XXXXXXXXX which are either at the top level normally or under a nest of numbered subdirectories The timekeys file contains version information the number of frames contained in each frame file and a map into the frame files The number of total frames in the frameset is si zeof timekeys 12 24 The metadata file is a frame file but rather than containing time centered data it contains data common to all frames in a trajectory The metadata file may contain an empty frame Typical fields in this file include TITLE and INVMASS The clickme file is an artifact of selecting files in a GUI browser like VMD The file browser won t allow a user to select a directory rather it clicks through to the underlying files Selecting the clickme file results in VMD actually selecting the enclosing directory Very large framesets 100 s of thousands of frames can exceed directory files storage limits so framesets can use a DeepDir hierarchical subdirectory structure to get around that limit The ddparams file contains two ASCII integers ndirl and ndir2 that describe a two level subdirector
144. hema for the vdw_12_6 nonbonded type name type description sigma FLOAT VdW radius in LENGTH epsilon FLOAT VdW energy in ENERGY The functional form is V a r 6 r where a and b are computed by applying either the combining rule from nonbonded_info or the value from nonbonded_combined_param to obtain o and e then computing aij 4eo and bij 4eo For both default and table the van der Waals contributions are in the nonbonded_vdw Hamiltonian category while the near electrostatic contributions are in nonbonded_elec force on1y contributes to no category and this is debatably a bug 7 3 2 Nonbonded tail corrections The truncation of van der Waals forces to a cutoff neglects the energy of the r term over the volume beyond r gt Reut This term decays as Reut and thus can be significant enough to warrant a tail correction term to the total energy of the system as well as an associated correction to the pressure The tail correction represents an averaged r interaction between particles outside of Reut from each other The term depends on the number of particles in the system the average dispersion and the current system volume the precise form depending on the means by which the term has been tapered none 2 Ur 2MyN 1 tail 3 YY BE e shift 4r N 1 Ural 3 YT BE cut e clswitchorc2switch _ 2m N 1 1 at x Ural Fy Ri 1 3 Jo E de
145. hival purposes translate adapt SOFTWARE engineering or otherwis supplied hereunder b or distribute the SOFTWARE or use the SOFTWARE time sharing benefit of third parties on the SOFTWARE d any applicable law without DESR information MATERIALS notify DESR use or disclosure of recognizes that r rule or materials other than as ES immediately the c export SOFTWARE sourc lease import or create or attempt to creat cod loan or otherwis or or regulation ES s prior written approval contained in or related to the SOFTWAR expressly authorized of any actual or imminent unaut and or any R the unauthorized use or disclosure of any o will give rise to irreparable injury to licensors for which monetary damages may agrees that DESRES its affiliates injunctive relief against the breach or obligations hereunder which may be available Derivat Source Code provided by D ESR ES to LIC in addition to any The ive ty to enter into this Agreement b Ce granted hereunder on the ENS will abide by and the terms and LIC T E he SOFTWARE ENS shall not only as necessary a modify EE LIC create derivative works from except in accordance with the Derivative Work Permissions set forth in this paragraph or any portion thereof t
146. home box e n is the number of processes along the X axis of the global cell e mg is the number of processes along the Y axis of the global cell e ng is the number of processes along the Z axis of the global cell By definition then ninong is the total number of Desmond processes Note The number of processes along each axis may be constrained by the requirements of the nonbonded terms if a discrete Fourier transform is used to implement Ewald summation see Nonbonded far interactions if not it outputs an error message and halts Assuming a homogeneous particle density throughout the global cell it s most efficient if the relative number of boxes along each axis is as close as possible to the relative proportions of the global cell so that each box is as close as possible to a cube This minimizes the surface to volume ratio of each box A surface represents a boundary between boxes so a minimal surface minimizes interprocessor communication For example for a relatively homogeneous system with dimensions 90 x 90 x 50 running on 32 processes a partition of n 4 ng 4 ng 2 is most efficient 42 Chapter 5 The Global Cell Desmond Users Guide Release 3 4 0 0 7 If you d like Desmond to set the number of processes assigned to a given axis then instead of setting it explicitly set the corresponding parameter to zero To allow Desmond to determine how to partition the global cell along all three
147. ial is purely TG harmonic integrated with velocity Verlet this quantity is sometimes of technical interest column E_f The eneseg plugin also reports pressure P volume V and temperature T as well as a temperature for each temperature group identified in the structure file T_N The header of the eneseq excerpted with a few columns below gives the number of particles N_at oms the number of degrees of freedom N_dof the total charge q_i and squared charge q_i together with other sometimes pertinent information 5dhfr production parameters Simulation started on Wed May 19 15 36 52 2010 sumi q_i 10 999998 sumi q i 2 7582 727781 N_atoms 23558 N dof 70674 70674 n_pressure_grp 23558 n_frozen_atoms 0 O time ps Lik kcal mol 2 E_p kcal mol 3 E_k kcal mol 0 000 7 24497189e 04 7 24497189e 04 0 00000000e 00 0 050 7 39533259e 04 8 30207059e 04 9 06738003e 03 Note Not all integration schemes have a conserved energy Details are discussed in Dynamics Note When performing initial velocity thermalization place this plugin before the eneseq plugin on the list of plugin names Table 4 14 Configuration for eneseq name description first First time for this action Time interval Time between actions Time name The output file name String 4 6 7 maeff_output The maeff_output plugin writes a structure file in the deprecated Maestro fil
148. iciently small Note Setting o oo makes erfc 1 and erf 0 so that the electrostatic potential is computed entirely as a cutoff pairwise interaction users operating in this regime should consider setting nonbonded far to none Because the nonbonded forces are partitioned into near van der Waals and electrostatic and far electrostatic com ponents they share a number of arguments in common such as the splitting parameter o and the near cutoff radius Rent Table 7 16 Configuration for nonbonded name description r_cut Distance at which near interactions vanish Length gt 0 n_zone Number of polynomial regions for potential interpolation functions Integer gt 0 sigma Electrostatic splitting parameter Length gt 0 near Configuration for the near nonbonded Can be set to none configuration far Configuration for the far nonbonded Can be set to none configuration 7 3 1 Near interactions force nonbonded near type default table force only taper none shift clswitch c2switch r_tap Rtap average_dispersion v optional nonbonded near type specifies the method used to compute nonbonded near interactions Some of these meth ods are built in and some are provided by extensions The built in ones e default van der Waals using a tuned Lennard Jones computational pipeline and an interpolating function for electrostatics e table an alternate implementatio
149. iety INTEGER moiety assignment nbtypeA INTEGER entry in nonbonded_param for A state nbtypeB INTEGER entry in nonbonded_param for B state chargeA FLOAT charge in the A state chargeB FLOAT charge in the B state chargeC FLOAT charge in the C state The chargeC column is optional It is only read if the alchemical configuration requires a third charge state 17 3 2 Bonded terms Alchemical bonded terms are to be treated by creating a table analogous to the non alchemical version but replacing each interaction parameter with an A and a B version An example is given in Schema for alchemical_stretch_harm records As a naming convention the string alchemical_ should be prepended to the name of the table 17 3 Alchemical systems 131 Desmond Users Guide Release 3 4 0 0 7 Table 17 11 Schema for alchemical_stretch_harm records name type description r0A FLOAT equilibrium separation in A state fcA FLOAT force constant in A state r0B FLOAT equilibrium separation in B state fcB FLOAT force constant in B state po INTEGER Ist particle pl INTEGER 2nd particle moiety INTEGER chemical group Alchemical harmonic stretch terms have a functional form given by interpolating between the parameters for states A and B 17 3 3 Constraint terms No support is offered for alchemical constraint terms at this time If particles A
150. ile for Viparr must contain particle bond and global_cell tables The particle table must contain at a minimum the anum column for atomic number resid resname chain and segid columns will also be used if provided to distinguish residues from each other See Appendix Appendix DMS file format for the specification of these columns and tables Viparr uses atomic numbers and bond structure graph isomorphism to match residues to templates Thus if you have nonstandard atom or residue PDB names you do not need to modify them and you do not need to be concerned about the atom and residue names used in the force field You can however modify atom and residue names for your own purposes if you wish In particular Viparr identifies the N and C terminus versions of the residues correctly as well as protonated and deprotonated versions of a residue even if you do not identify them as such 6 2 2 Running Viparr Once you have a complete structure in DMS format use Viparr to add forcefield information The command line for running Viparr is viparr input dms output dms d ffdir x f ffname Here f fdir is path to a forcefield directory and f fname is the the subdirectory of SVIPARR_FFDIR containing a forcefield directory The available force fields are listed in Force fields built into Viparr Multiple forcefields can be provided this allows one for example to use either tip3p or tip4p with the charmm27 forcefield by specifying f c
151. ile can now be input to vrun which analyzes the results according to the manner specified in the configuration For example you can specify that vrun print the energy of the system for each frame or the forces on each particle at each frame Other tools such as VMD a freely available visualization application can be used to analyze results in addition to or instead of vrun 3 8 5 Customizing Desmond Desmond modularizes its functionality in the form of extensions An extension is a software module that implements a discrete set of capabilities compiled separately so that it can be added to or removed from an existing application The capabilities are further divided logically into units of functionality called plugins As it runs the Desmond executable calls plugins as specified in the configuration file for its application In this way you can execute the functions that you need while skipping those that you don t Each Desmond application has a main loop which it repeats one step in the minimization process one simulation timestep or one trajectory frame loaded Plugins can be called during this loop to perform their work repeatedly as the simulation unfolds For example the plugin eneseq computes system energy temperatures pressures and other data breaking down the energy into various categories then writes the result to the specified output file For example randomize_velocities reinitializes the velocities of the particles in
152. in parallel and describes how to configure Desmond applications and built in plugins as well as the optional profiling mechanism 4 1 About configuration Desmond reads configuration parameters from a configuration file specified on the command line The simplest way to configure a simulation is to copy one of the sample configuration files provided and edit it See the README txt file for the location of these files For those who wish to edit extensively or create their own configuration file syntax is described in Appendix Configuration syntax Configuration files are divided into sections with the configuration information for a given application going into the section named for that application In addition other sections configure other aspects of the simulation such as the global cell the force field constraints if any and the integrator The same configuration file can apply to any Desmond application Note Schrodinger s release of Desmond uses a simplified configuration file format described in Appendix C of their Desmond User Manual that differs from Desmond native form described below This simplified configuration file is converted to the native form whenever the user executes SCHRODINGER desmond Configuration file sections are app mdsim remd minimize vrun boot file p the structure file global_cell force migration integrator profil
153. interactions are computed Table 18 3 Particle properties obtained from ffio sites block Site property ffio sites attribute particle type ATOM or PSEUDO ffio_type charge units of e ffio_charge mass atomic units ffio_mass van der Waals type string key ffio_vdwtype Note The value of i0_comb_rule must be same for all ct blocks 18 1 Format 135 Desmond Users Guide Release 3 4 0 0 7 All other interactions are determined by additional subsections of the fio_ ff block For example two body stretch harmonic stretch terms are found in a subblock called fio_bonds and van der Waals interactions are specified by the VDW type and by a subblock called fio_vdwtypes Note Because the Maestro file format is designed to be extensible many other interaction types are possible consult the documentation for the specific force terms you wish to employ to determine which structure file records contribute to those terms Note The Maestro file is sometimes referred to as a MaeFF file when it has force field parameter assignments present The Maestro Desmond system builder tool will output MaeFF files with the file name suffix cms Note The DMS file can not be directly converted into a MAE file A workaround is to use VMD to convert a DMS file to a MAE file This conversion will not include force field parameters present in the DMS file however Force field parameters c
154. ion of the format of the structure file and is other wise not used The unnamed block is followed by one or more connection tables These are called _m_ct blocks or simply ct blocks met 4d s_m title r chorus_box_ax r_chorus_box_ay r_chorus_box_az this is the title 25 0 0 0 0 0 m_atom 2 i_m_mmod_type r_m_x_coord r_m_y_coord r m z coord 133 Desmond Users Guide Release 3 4 0 0 7 1 0 326 0 704 0 726 2 1 0 431 1 245 1 295 The ct block in the previous Example shows four attributes plus an array block called m_atom The attributes are m_title chorus_box_ax chorus_box_ay and chorus_box_az The array block called m_at om has three attributes and two records The attribute names are prepended by s_ r_ or i_ depending on whether the corresponding value is a string text real number or integer respectively Note In the discussion below these prefixes are ordinarily excluded Attributes names also encode the owner of the attribute that is the name of the application responsible for managing that quantity For example the attribute name prefix m_ indicates that Maestro is responsible for managing that attribute In an array each record has a one based index followed by values for the attributes of the block one for each record The size of the array block is given by the number in square brackets after the name In the Example above the value corresponding to chorus_box_ax is 25 0 and
155. ion parameters for each of these applications are described below 4 4 1 mdsim mds im is Desmond s main molecular dynamics simulation code It s configured as shown in mdsim title last_time t plugin f checkpt 1 W Table 4 2 Configuration for mdsim name description title A short string to include in various output files by default no title string last_time Time at which to stop the simulation in picoseconds relative to the reference time given as part of the global cell configuration see Configuration time plugin Description of the main loop plugins to call during simulation See Using plugins configuration checkpt Checkpoint configuration See Checkpointing configuration Checkpointing Because mds im can run for a long time during which hardware can fail checkpointing allows you to restart a simu lation from a backup file called a checkpoint A checkpoint file is a snapshot of the entire state of the computation and can therefore be quite a large file However because their purpose is to restart an interrupted simulation checkpoint files can be discarded after the simulation completes Desmond checkpoints are designed such that the state of a sim ulation restarted from checkpoint is bitwise identical to the state of simulation at the point when the checkpoint file is written Configuration information for checkpointing appears as shown in
156. ional by default empty List of names tol Relative tolerance for the constraint algorithm Real gt 0 use_Reich employ Reich s rigid motion constraint algorithm for HOH constraints Optional by default true Boolean use_reshake Compensate for double to single precision rounding effects Optional by default true Boolean 66 Chapter 8 Constraints Desmond Users Guide Release 3 4 0 0 7 8 1 Single precision resolution and constraints The degree to which a set of distance constraints can possibly be satisfied is a function of the resolution of the rep resentation of particle positions When the atomic coordinates are represented by single precision numbers there is some possibility that numerical errors coming from constraints with poor position resolution can accumulate during the course of the simulation Particle positions are represented in a local coordinate system whose origin depends on the owning process The dimensions of that local cell are proportional to the distances along Cartesian axes between representable positions in real space and thus inversely proportional to resolution Thus when the dimensions of the local cell increase by running on larger systems or with fewer processors the resolution decreases Time resolution is also relevant Clearly the more time steps used for a given simulated time the more space resolution errors accumulate but empirically the relationship is no
157. ions The structure file contains all particle and bond information but has as yet no information about the force field describing the interactions between particles b To add the force field information the structure file is input to Viparr You specify the force field you wish to use and Viparr outputs a structure file with the force field informa tion added It can access a set of databases specifying the required force terms for the various molecules in the chemical system Viparr reads the structure file and appends the necessary force terms in a separate section of the file You now have a structure file that defines the particles and forces in your simulation c If you wish to use constraints in your simulation you now run build_constraints By default the script constrains the bond length of all bonds involving hydrogen atoms as well as the angle in all water molecules The out put is a new structure file with the constraint terms added You now have a structure file that describes the particles and forces in your simulation as well as any constraints you wish to apply 2 The simulation still needs to be configured which involves specifying the values of parameters in a configuration file The simplest way is to start with an existing con figuration file and edit it Running Desmond provides an overview of configuring the simulation For details about specific configuration file parameters see the chapters that discuss the
158. ions before turning off Lennard Jones interactions An example of a sensible A schedule for a binding free energy simulation is given in weights vdw 0 00 0 25 0 50 0 75 1 00 1 00 1 00 1 00 1 00 es 0 00 0 00 0 00 0 00 0 00 0 25 0 50 0 75 1 00 To carry out ligand binding free energy simulations you must specify which atoms in the system belong to the ligand by setting grp_ligand for these atoms to 1 and for all other atoms to 0 in the structure file Table 10 2 Configuration for binding FEP name description weights vdw parameterizes intermediate Lennard Jones interactions List of 0 lt Reals lt 1 weights es parameterizes intermediate electrostatic interactions List of 0 lt Reals lt 1 10 1 3 Alchemical free energy simulations Alchemical free energy simulations are configured as shown in force term x keyx type alchemical weights bondA bj bij bf bondB by bry bT vdwA v Un v vdwB ud Wi fe ate v qA le a orn ef aB ice Clin TE cP qc cS GYA bai co optional force nonbonded near type alchemical softcore Same parameters as default In alchemical free energy simulations a part of the system called A is changed into something else called B In this transformation some atoms change their Lennard Jones parameters and partial charges and some bonded interactions change their parameters We introduce a fami
159. ist of strings At each application the pprofile plugin divides the simulation cell into a number of slabs parallel to the z axis Contributions to the pressure from particles located within each slab are computed where each particle s position is wrapped to the central global cell These values are output to a frameset The time between pressure profile calculations can be specified in addition the time between profile output can be given separately in which case the average of the values collected over the preceeding interval will be written Output frameset contains the following fields FORMAT the string PPROFILE_V1 CHEMICALTIME the simulation time at which the data was written NSLABS the number N of partitions of the simulation cell NEVALS the number of virial evaluations that have been averaged to compute the data in the frame UNITCELL the global cell dimensions at the current time CORRECTION the zx y and z diagonal components of the long range correction to the pressure from the nonbonded tail correction see Nonbonded tail corrections kinetic 3N doubles listing the x y and z diagonal components of the pressure for each slab due to particle kinetic energy C 3N doubles listing the x y and z diagonal components of the pressure for each slab due to interactions in Hamiltonian category C total 3N doubles listing the x y and z diagonal components of the pressure for
160. itten The syntax of the meta keyword is meta meta_acc height_width collective_vars where meta_acc is an integer that references a member of the set of metadynamics accumulators height_width is an array of height and widths to use for newly inserted kernels and collective_vars is an expression for the collective variables The length of the collective variables array is equal to the dimension of the accumulator and the length of the height_width array is one more than the dimension of the accumulator The height is the value of the kernel at its center The height_width array is only evaluated when a kernel is added to the potential 11 4 2 Metadynamics example An example configuration for a simple metadynamics simulation is given below This configuration file biases the inter atomic distance of the atoms given by GIDs 0 and 1 11 4 Metadynamics 105 Desmond Users Guide Release 3 4 0 0 7 define the accumulator declare_meta dimension 1 only one collective variable cutoff 9 in units of widths first 0 0 begin dropping immediately interval 0 200 wait 0 2 picoseconds between drops nam kerseq log kernels to kerseg initial gt no initial kernel file p atomsel index 0 1 meta 0 use accumulator 0 array 0 2 0 1 height is 0 020 kcal mole width is 0 1 A dist p 1 p 0 coordinate is distance between atoms 0 and 1 More examples of metadynamics can be found in
161. iven later terms can have affects on previous terms The KEYPATH term expands to a key which can be extended by a list number or table key indexes New lists or tables are created when necessary to accommodate these indexes The KEYPATH term resolves to a possibly newly created subtree of the configuration An assignment KEYPATH VALUE obliterates the previous subtree if any and replaces it with the expansion of the VALUE term A list KEYPATH term can be extended with an assignment of the form KEYPATH VALUE An enclosure KEYPATH CONFIG changes the subtree to an empty table unless it is already a table and merges the two tables by appending to the subtree s table the KEYVAL terms of the CONFIG This is analogous to the behavior of namespaces in C An INCLUDE term expands into the text of the file indicated by the QATOM in its production rule with comments removed This text is presumed to be a CONFIG term and its sequences of KEYVAL terms are inserted into the stream of terms in which the text is included Inclusion is properly nested an included file can include other files referring to it by paths relative to itself The include FILE command line option is equivalent to cfg include FILE Ark types summarizes the information above Table 15 1 Ark types value type description atom A string bare or quoted any of the three standard quot
162. ixed block For this purpose if the consituent elements both represent a continuous thermostatting scheme Langevin or NoseHoover the effective relaxation time is the harmonic mean of their individual relaxation times If instead the Antithetic thermostat is one of the elements then the relaxation time is that of the other element To have no thermostat set thermostat none 88 Chapter 9 Dynamics Desmond Users Guide Release 3 4 0 0 7 The only type of barostat supported is MTK to have no barostat set barostat none The MTK type introduces extended variables s and 7 as described in Piston_NPH constant pressure and enthalpy and add an extended variable energy equal to gt 1 2W Po Tr TB s B s The MTK barostat s velocities 1 can be thermostated by one of Langevin NoseHoover or Antithetic to have no thermostat set barostat thermostat none Without a thermostat the MTK barostat steps evolve the r p s 7 variables as a Martyna Tobias Klein barostat without its Nos Hoover chain 7 A m Fi W 5 nisi W Bi 1 1 Ng A n pi W jaa eer B s Po B s B s K Za me With the Langevin thermostat the barostat steps evolve the r p s 7 variables as a Langevin piston The barostat mass W is given by W 3N d kpTy nptp F A n ri W i M8 W Bi 1 1 Ng A m p W a Pn Be Po B s 1B 6 y D 2 1 098 0 mam where each of the components of th
163. l force field this means that the interaction between the ligand and the rest of the system consists of nonbonded van der Waals and electrostatic interactions only Desmond uses the following form for the interaction potential Vs aoe qidj Vtrnts XO falli rleo A iEL jES i i jes Ir 75 10 1 Configuring free energy simulations 95 Desmond Users Guide Release 3 4 0 0 7 where f is the following softcore potential governed by parameter a g 2 g Bama e a v 20 a l v 20 von where and o are the usual Lennard Jones parameters The soft core potential is used so that the energy difference W w is always bounded for v 0 even when non ligand atoms are infinitesimally close to the ligand atoms In theory the path of changing v C from 0 0 to 1 1 should not affect the computed AF because free energy is a state variable independent of history and determined only by the thermodynamic state Practically however the choice of the v C path affects both the convergence and the stability of simulations Most importantly when the ligand and the rest of the system interact through the softcore potential that is v 4 1 non ligand atoms can overlap with ligand atoms in space causing the Coulombic interaction between their partial charges to diverge unless this electrostatic interaction has been turned off that is C 0 Hence it is always necessary to turn off Coulombic interact
164. lated in discrete timesteps like the frames of a film From one step to the next a tiny slice of time goes by and atom positions update accordingly Atoms move time advances atoms move again Frame by frame the simulation builds a movie for example a microsecond in the life of a protein How the atoms move in which direction by how much is determined by e the initial atom positions and velocities e the thermodynamic environment and e a molecular mechanics force field The molecular mechanics force field is a set of functions and parameters that describe the potential energy of the interactions between the particles in a chemical system In addition to its position each particle has an associated charge and atomic number as well as a list of the bonds that it participates in Using this information the force field models the forces exerted on each particle by every other particle thus determining each particle s acceleration Simulations such as Desmond s that use the laws of classical physics can only approximate full quantum mechanical reality They bow to the limits of computer performance solving the full set of quantum mechanical equations would take far too long Though merely an approximation integrating Newton s laws of motion for so many particles still means a great many computations for each step forward Molecular dynamics simulations therefore face a dilemma For accurate results the simulation timestep must
165. lds have the property that the dynamics they produce has no net center of mass translation Hence the degrees of freedom of the system are effectively reduced by 3 The flag ignore_com_dofs causes 3 to be subtracted from any appropriate degree of freedom counters within Desmond such counters are used by some integrators and by some output diagnostics Changing this flag would for example alter reported simula tion temperatures obtained by dividing kinetic energy by degrees of freedom For large systems its effects become negligible Table 7 1 Configuration for force name description bonded exclude bonded terms to turn off Optional by default empty List of names bonded include bonded terms which must be turned on overrides exclude Optional by default empty List of names virtual exclude virtual terms to turn off Optional by default empty List of names virtual include virtual terms which must be turned on overrides exclude Optional by default empty List of names constraint exclude constraint terms to turn off Optional by default empty List of names constraint include constraint terms which must be turned on overrides exclude Optional by default empty List of names polar exclude polarization terms to turn off Optional by default empty List of names polar include polarization terms which must be turned on overrides exclude Optional
166. like the regular maximum gibbs_min Class Normal Arguments e T scaling temperature e a array Return T log O exp a T A softened minimum is infinitely differentiable unlike the regular minimum helix Class Normal Arguments e tol tolerance e gids list of particles 5x length of phipsis 2 e phipsis list of w pairs of dihedral angles in radians 148 Chapter 19 Enhanced sampling function reference Desmond Users Guide Release 3 4 0 0 7 Return Returns the count of how many groups of particles from gids taken 5 at a time have the property that the dihedral angle of the first 4 is within tol of the corresponding phi and the dihedral angle of the last 4 is within to of the corresponding psi i e for the group x1 x2 x3 x4 x5 and angles 4 dihedral_angle x1 x2 13 x4 lt tol and dihedral_angle x2 13 x4 15 lt tol length Class Normal Arguments e a array Return the number of elements in a log Class Threaded Arguments a array Return the element wise logarithm of a mass Class Threaded Arguments e a array of gids Return the element wise mass of a meta Class Special Form Arguments e mid integer index of a metadynamics accumulator zero indexed e array of the gaussian height followed by the gaussian widths e array of the collective variables Return the metadynamics potential at the current location in the collective variables Note that the height
167. lly specified behavior with respect to periodic images and the user should review this behavior to ensure that the correct periodic images are chosen 104 Chapter 11 Enhanced Sampling and Umbrella Sampling Desmond Users Guide Release 3 4 0 0 7 11 4 Metadynamics Metadynamics is a free energy perturbation method which enhances sampling of the underlying free energy space by biasing against previously visited values of user specified collective variables The biasing is achieved by dropping kernels only Gaussian kernels have been implemented at the current location of the simulation in the phase space of the collective variables This history dependent potential encourages the system to explore new values of the collective variables and the accumulation of potential allows the system to cross potential barriers much more quickly than would occur in standard dynamics 11 4 1 Usage The enhanced sampling plugin implements metadynamics by using the declare_meta header to define the accu mulator for the history dependent potential and using the meta function to compute the potential for the interpreter Each call to declare_meta creates an independent kernel accumulator which does not share kernels with any other accumulator The accumulators are indexed in the order that they are declared The parameters to declare_meta are as follows dimension defines the dimension of the collective variable space which must be a positive integer
168. lude interactions and other molecules we believe that for simulations where water is the solvent running at typical temperatures Influence of finite precision resolution and timestep on energy drift captures the ballpark drift contribution one can expect to see from constraint resolution issues Table 8 4 Influence of finite precision resolution and timestep on energy drift step size 10A 20A 30A 40 A 60 A 80 A 2 0 fs 0 08 0 44 0 01 0 19 0 03 0 13 0 04 0 10 0 04 0 11 _0 02 0 07 1 0 fs 0 25 0 47 0 01 0 27 0 090 17 0 07 0 14 0 46 0 14 1 22 0 07 0 5 fs 0 10 0 55 0 57 0 28 1 97 0 26 4 58 0 62 14 21 0 48 31 06 0 62 0 25 fs 1 7 2 3 143Q 3 420G3 2 80 6 1 3 166 6 2 1 238 6 3 0 8 1 Single precision resolution and constraints 67 Desmond Users Guide Release 3 4 0 0 7 68 Chapter 8 Constraints CHAPTER NINE DYNAMICS This chapter summarizes the basic concepts of particle dynamics and describes how to configure the migration interval timestep scheduling pressure and temperature It also describes each of the available dynamical systems and how to configure them 9 1 Particles and mechanics Molecular systems are collections of particles evolved by some variant of the dynamics of Newtonian mechanics Newtonian mechanics can be summarized by a few conserved quantities usually a scalar with units of energy and a probability density Certain variations to the eq
169. ly of interpolating potential functions parameterized by A 96 Chapter 10 Free Energy Simulations Desmond Users Guide Release 3 4 0 0 7 and b4 bP v4 vP c4 cP c0 The potential function of H is the sum of electrostatic softcore Lennard Jones and bonded terms Va r V r VY r VPM The interpolating electrostatic interaction is computed using partial charges linearly interpolated between A and B and C if a qC schedule is given In other words it is computed using the charges A A B B qi cS q fa AO The alchemical charges q and q and q if a qC schedule is given are taken from the structure file The Lennard Jones interactions for a pair of atoms and 7 changing their combined Lennard Jones parameters from ef o to cB oF Ei 7 the following intermediate potential is used gt gt gt A A VAF T foa Ili Fill es 0 for llFi Fill eg og where f is the softcore potential defined in Equation 10 1 The intermediate bonded interactions are the linear interpolations between the interactions with parameters in A and B VEA VIA BEV BOM r where the A state and B state bonded interactions V pone and pa are taken from the structure file Ar guably alchemical partial 14 terms see the partial 14 description lt descr partiall4 gt should transform accord ing to the vt4 P cl4 B and adopt the soft core functional form of Equation 10 1 Within a DMS file
170. ly update an output file for example eneseq compute_forces energy_groups and gibbs output can use an empty string as a filename in this case data is written to the standard output However maeff_ output and checkpt require real file names 4 6 Configuring the built in plugins Desmond is compiled with various plugins which are therefore available to all Desmond applications These plugins offer a range of commonly useful functionality configuration information for them all is discussed below 4 6 1 anneal The Desmond anneal plugin periodically updates the temperature setting of the anneal integrator during an mdsim run The anneal integrator is actually a thin shell around any other Desmond integrator Hence there are two places in the configuration that need to be changed in order to use the anneal plugin Integrator setup The integrator section of the configuration normally has the following form integrator type name name integrator specific options other non specific integrator options In order to enable the anneal plugin the above should be changed as follows 4 5 Naming output files 25 Desmond Users Guide Release 3 4 0 0 7 integrator type anneal anneal type nam name integrator specific options other integrator options This wraps whatever integrator symbol was asking for inside the anneal integrator and thereby makes it responsive t
171. m is simply the interaction of a Gaussian cloud with itself The second term which is only relevant for systems that are not charge neutral is an additional interaction between a Gaussian cloud and a uniform background charge of density p q V This background charge in non neutral simulations is required to cancel the non zero contributions from the system charges which would otherwise cause the Ewald sum to blow up 7 4 4 Virtual sites Virtual sites a form of pseudoparticle are additional off atom interaction sites that can be added to a molecular system These sites can have charge or van der Waals parameters associated with them they are usually massless The TIP4P and TIPSP water models are examples that contain one and two off atom virtual sites respectively Because these sites are massless it is necessary to redistribute any forces acting on them to the particles used in their construction A consistent way to do this can be found in Gun 1984 The virial in most cases must also be modified after redistributing the virtual site force The types of virtual site placement routines are described below Ic2 virtual site The 1c2 virtual site is placed some fraction a along the vector between two particles i j Ty 1 cri cir Table 7 22 Schema for virtual_lc2 records name type description cl FLOAT coefficient 1 po INTEGER pseudoparticle id pl INTEGER parent atom i p2 INTEGER paren
172. m the file assuming it exists To access frames there are two methods Reader frames returns a Framelter object for iteration over frames Framelter has two methods the usual next method which returns a Frame and skip n 1 which advances the iterator by n frames without necessarily reading anything Framelter is a very poor iterator once a frame has been read or skipped it can t be loaded again you have use a brand new Reader Reader frame n returns the nth frame 0 based index Currently only the dtr plugin supports this method e Writer Writers are initialized with a path and either an array of Atoms or an atom count If the Writer supports structure writing Atoms must be provided if the Writer only writes frames either one will do If the writer supports frame writing Writer frame f appends frame f to the end of the file Writer close will be invoked when the Writer goes out of scope but it s not a bad idea to invoke it explicitly Finally there is a guess_filetype path default None function in the mol file module which returns a Plugin based on the file name or the default if none found 13 4 Python tools for trajectories and framesets 117 Desmond Users Guide Release 3 4 0 0 7 Table 13 3 molfile object properties and methods property or method description Atom altloc PDB altloc value Atom anum atomic number Atom bfactor temperature factor Atom
173. mbined arbitrarily desmond include config_file cfg last_time 10 0 which is equivalent to the following configuration text title this is an example pi 3 14159 file myDoc txt matrix 1 0 0 0 1 0 0 O 1 J options verbose yes Nsteps 100 vec 1 2 3 last_time 10 0 Repeated key assignments override previous ones In the following Example both assignments have the effect of producing the configuration X 2 desmond cfg X 1 X 2 desmond cfg X 1 cfg X 2 Through keypaths elements of a configuration can be individually overridden from the command line desmond include config_file cfg matrix 2 1 1 1 options verbose no which results in a configuration equivalent to title this is an example pi 3 14159 file myDoc2 txt matrix 1 0 0 0 1 0 1 1 1 options verbose no Nsteps 100 vec 1 2 3 The enclosure syntax can be used to extend a table 15 1 Examples 123 Desmond Users Guide Release 3 4 0 0 7 desmond include config_file cfg options verbose no Nsteps 50 which results in a configuration equivalent to title this is an example pi 3 14159 file myDoc2 txt matrix 1 0 0 0 1 0 1 1 1 options verbose no Nsteps 50 vec 1 2 3 Conversely an assignment such as desmond include config_file cfg options verbose no Nsteps 50 results in the configuration title this is an exampl
174. me convention as in atommaps i_fepio_aj The second atom in the dihedral in orignal_ct Negative atom number can appear here by the same convention as in atommaps i_fepio_ak The third atom in the dihedral in orignal_ct Negative atom number can appear here by the same convention as in atommaps i_fepio_al The fourth atom in the dihedral in orignal_ct Negative atom number can appear here by the same convention as in atommaps 18 2 Preparing the structure file for Free Energy Simulations 139 Desmond Users Guide Release 3 4 0 0 7 fepio_exclmaps This indexed block maps the exclusions from original_ct onto perturbed_ct Table 18 10 fepio_exclmaps properties property name description 1 fepio_ti Exclusion number in original_ct Negative exclusion number indicates that this exclusion does not exist in the original ct and it involves at least one dummy atom 1 fepio_tj Exclusion number in perturbed_ct Negative exclusion number indicates that this exclusion does not exist in the perturbed ct and it involves at least one dummy atom If both i_fepio_ti and i_fepio_tj are 1 this exclusion does not exist in either the original or perturbed ct and is an extra exclusion to prevent dummy atoms in original_ct from interacting with dummy atoms in perturbed_ct 1 fepio_ai The first atom in the exclusion in original_ct Negative atom numbers can appear here by the same convention as in atommaps 1 fepio
175. merically solves at every timestep thus computing a new position and velocity for every particle in the system The differential equation is based on the laws of Newtonian mechanics applied to particles in the system but modeling some physical systems requires augmenting the differential equations Desmond implements three broad categories e Ordinary differential equations that hold certain measures constant Verlet constant volume and energy Nos Hoover constant volume and temperature MTK constant pressure and temperature and Piston constant en thalpy e Stochastic differential equations that hold certain measures constant and in which one or more of the terms is a stochastic process Langevin constant volume and temperature and Langevin constant pressure and tempera ture e Ordinary differential equations coupled to feedback control systems that keep a certain measure within a certain range Berendsen constant temperature and Berendsen constant temperature and pressure The particular algorithm that Desmond uses to solves the differential equation is called the integrator Integrators are described in detail in ntegrator Desmond allows you to specify other aspects of the motion in your simulation as well For example if you re using certain integrators you may wish to remove the center of mass motion of the chemical system Even with optimizations such as the Fourier space computation far interactions are expensive to compute
176. metere describing how a pseudoparticle moves relative to its parent atoms Table 17 7 Local interaction metatables name type description name TEXT name of the table for an interaction form Schema for the bond_term constraint_term virtual_term and polar_term tables described in Metatables for local particle interactions Each table name corresponding to the values in the local term metatables is the designated string for a particular functional form The required columns for these tables is given in the next section Note that creators of DMS files are free to implement the schema as an SQL view rather than as a pure table a reader of a DMS file should not assume anything about how the columns in the table name have been assembled 17 2 2 Nonbonded interactions The functional form for nonbonded interactions as well as the tables containing the interaction parameters and type assignments are given by the fields in the nonbonded_info table shown in Schema for the nonbonded_info table Table 17 8 Schema for the nonbonded_info table name type description name TEXT nonbonded functional form rule TEXT combining rule for nonbonded parameters There should be exactly one record in the nonbonded_info table Like the local interaction tables described by ref tab localterm the name field indicates the functional form of the nonbonded interaction type If the particles do n
177. mics are given by the ODE fi pi mi u jui Bi VU r Bek JO Y Wpill mi Ci viv w ilx i j v y Jwi CO vlvl 1 0 vi vi u_ Ci II where 10 C1 7 with ol kgT N and C kgT where N is the number of degrees of freedom of the governed particles j Recall from Tenner ature that x i denotes the thermostat which governs particle 2 This system preserves the scalar H r C p v Ho r p 20 2w ee and the phase space density Q exp Y kaeT YC aca 0 J i ij In particular if T Tk T then the density Q exp kgT Ho r p De 2w adri 2 is preserved Hence the trajectories of these equations if ergodic sample r p from the canonical ensemble with temperature T The current numerical implementation of the ODE updates each Nos Hoover chain as a separate step from the gov erned position and momentum variable updates Because these updates are inexpensive they can be done multiple times m with a smaller timestep proportionate to 1 m for higher accuracy In practice we usually set m 2 Table 9 5 Configuration for NH_NVT name description thermostat mts The number of discrete updates within the chain Integer gt 0 thermostat tau The time constants determining the length and masses of the chain variables List of Time gt 0 9 6 3 Anti_NVT Antithetic constant volume and temperature The Ant i_NVT dynamical
178. mics are specified in a section named integrator in which one specifies the conditions for evolving the system forward in time The integrator is configured as shown in integrator dt Ot respa 1 pressure temperature type key dynamics type key 1 specific options The type value specifies the dynamical system defining the system s evolution see Dynamical systems Additionally the type value is used as a key for any additional parameters that the selected system requires dz is the amount of simulated time between particle position updates Every position update is F t 54 F t Pilt 64 2 64 mj followed by a modification to account for any constraints see Constraints Because Desmond supports multiple timestepping the full timestep A between successive simulation steps might not be 6 but instead some integer multiple of it 6 is sometimes called the inner timestep and A n is called the outer timestep 70 Chapter 9 Dynamics Desmond Users Guide Release 3 4 0 0 7 Table 9 1 Configuration for integrate name description dt The time length of a position update step Time gt 0 respa breakdown of the integrator timestep configuration pressure configuration of the system pressure configuration temperature configuration of the system temperature configuration type Type of dynamical system to integrate Symbol
179. more strict limitation 3 Retone lt 4 L because clone regions cannot overlap Finally if a dimension 7 has not been partitioned at all 1 Relone lt qui These restrictions have been phrased in terms of a Cartesian global cell For a triclinic cell the concerns are analogous though the mathematical conditions more difficult to summarize 126 Chapter 16 Appendix Clone Radius Restrictions CHAPTER SEVENTEEN APPENDIX DMS FILE FORMAT All data in a DMS file lives in a flat list of two dimensional tables Each table has a unique name Columns in the tables have a name a datatype and several other attributes most importantly whether or not the column is the primary key for the table Rows in the tables hold a value for each of the columns Table names column names and datatypes are case preserving but case insensitive thus pArTiCLE is the same table as particle and NAME is the same column as name Of the five datatypes available in SQLite DMS uses three INTEGER a signed 64 bit int FLOAT a 64 bit IEEE floating point number and TEXT a UTF8 string In addition any value other than a primary key can be NULL indicating that no value is stored for that row and column A NULL value is allowed in the DMS file but might be regarded as an invalid value by a particular application for example Desmond makes no use of the atomic number column in the particle table but Viparr requires it Because DMS i
180. n Time interval Time between actions Time seed Seed for the random number generator Integer temperature The target temperature temperature 4 6 11 remove_com_motion The remove_com_motion plugin periodically removes net center of mess motion from the system velocities Configuration is shown in app plugin list key key type remove_com_motion first tf interval ti Although most systems in their exact mathematics have no net center of mass motion numerical implementations might have nonzero motion Most dynamical systems do not explicitly remove center of mass motion This plugin will periodically subtract off any net center of mass motion from the system 4 6 Configuring the built in plugins 35 Desmond Users Guide Release 3 4 0 0 7 Table 4 19 Configuration for re move_com_motion name description first First time for this action Time interval Time between actions Time 4 6 12 trajectory The trajectory plugin writes trajectory data using current simulation coordinates It is configured as shown in the Synopsis app plugin list key key type trajectory first tf interval ti name d write_velocity by mode m periodicfix bp center Ci Cm glue Y gn 1 write_last_step b Data is written as a set of frames in the directory d following output file naming conventi
181. n and or other materials provided with the distribution 3 The name of the author may not be used to endorse or promote products derived from this software without specific prior written permission HIS SOFTWARE IS PROVIDED BY THE AUTHOR AS IS AND ANY EXPRESS OR PLIED WARRANTIES INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES F MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT INDIRECT NCIDENTAL SPECIAL EXEMPLARY OR CONSEQUENTIAL DAMAGES INCLUDING BUT O sIMITED TO PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES LOSS OF USE ATA OR PROFITS OR BUSINESS INTERRUPTION HOWEVER CAUSED AND ON ANY HEORY OF LIABILITY WHETHER IN CONTRACT STRIC LIABILITY OR TORT INCLUDING NEGLIGENCE OR OTHERWISE ARISING IN ANY WAY OUT OF THE USE OF HIS SOFTWARE EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE T O QAWHH OHH zZ 20 2 4 V8 project s double to ASCII conversion library Copyright 2006 2011 the V8 project authors All rights reserved Redistribution and use in source and binary forms with or without modification are permitted provided that the following conditions are m
182. n attribute fieldname Boolean __knowsType__ typename true iff frame understands named C type Boolean _ framesize_ Number of bytes required for serialization __serialize_ string serialization __ type_ fieldname C type name for this field string _ count__ fieldname Number of elements in this field __ elementsize_ fieldname Size of individual elements in the named field __nbytes_ fieldname Number of bytes used to hold a field 13 4 2 generictrajectory module for simplified access The generic trajectory module provides an interface to the unified trajectory reader object We use this reader object to open any of the variety of trajectory files and trajectory like files that are supported by Desmond The module provides Trajectory and Frame types Not all frameset based trajectories have convenient POSITION and VELOCITY fields Frames with the FORMAT field set to WRAPPED_V_2 for DBL_WRAPPED_V2 will have those fields but frames in other formats will not To simplify access the generictrajectory module understands all Desmond formats and auto converts frames into Python objects with position velocity box title and time attributes Position and velocity are numpy arrays of double natoms 3 box is double 3 3 title is a Python string and time is a Python float C double The generictrajectory frames are similar to the full frameset frames above but have only the fixed attributes listed These Fram
183. n of width s Because the charge spreading and force and energy accumulation steps are done in real space with a localized but not compactly supported function a cutoff Rspread is used to truncate the Gaussian to zero Experiments have shown that spreading the charge onto more than 250 mesh points does not significantly improve accuracy Thus Rspreaa 18 typically selected to contain a sphere of approximately 250 mesh points for example 1 250 Y Rspread h 5 where h is the smallest mesh spacing along any axis The value of is then chosen such that erfe r y2 0 is small at the radius Rspread Table 7 21 Configuration for gse name description sigma_s Bandwidth parameter for Gaussian charge density interpolation 0 lt Length lt a V2 r_spread Cutoff parameter for Gaussian charge density interpolation Length gt 0 For more information see Sha 2005 Note Normally the GSE is not used in Desmond simulations 62 Chapter 7 Calculating Force and Energy Desmond Users Guide Release 3 4 0 0 7 7 4 3 Electrostatic self energy correction The Gaussian spreading of point charges creates non physical self interaction energies where a point charge interacts with itself To remove these contributions a self energy correction is added to the potential la a Uself a y 57 where q2 q and q q with the sums taken over all the particles in the system The first ter
184. n providing greater flexibility at the expense of performance by using interpo lating functions for both e force only similar to the default but without energy evaluations Provides increased performance Where energies do not need to be calculated you can achieve significantly greater performance by using the force only form of the nonbonded interaction This form of the nonbonded interaction cannot be used where energies need to be evaluated as with the energy_groups plugin table is an alternative to default which employs an interpolation scheme for both van der Waals and electrostatic computations This allows a tapering method to be applied to all near nonbonded interactions This is computationally slower 58 Chapter 7 Calculating Force and Energy Desmond Users Guide Release 3 4 0 0 7 The Lennard Jones 12 6 potential between two particles is Visa 3 oe T Yi where rj r r is the distance between two particles i and j Coefficients a and b depend on the types of the particles 7 and 7 Desmond reads per particle van der Waals properties a and b for particle 7 and constructs a and bij by a function of the per particle coefficients called a combining rule specified in the structure file Each near nonbonded potential function electrostatic or van der Waals 7 is truncated by a cutoff to 0 for r gt Reut If Reut is selected too aggressively Reut lt 9 then discontinuities in at
185. n2 n1 n3 and n1 n4 Note that it is important to choose particles for the glue that are as close as possible to each other in the input structure After the bond fixing step the centering step is performed if any particles in the center configuration have been specified A single translation is applied to the entire system in order to bring the geometric center of the center atoms to the origin The last transformation applied to the coordinates assuming periodicfix is enabled is to take connected sets of atoms and translate each set as a group so that their geometric centers are located within the central unit cell Here again the definition of connected use both the input structure bonds as well as bonds supplied by glue 36 Chapter 4 Running Desmond Desmond Users Guide Release 3 4 0 0 7 Table 4 20 Configuration for trajectory name description first First time for this action Time interval Time between actions Time name The output directory name for the frameset String write velocity Whether to include velocity information in output frames Boolean mode Open mode for the frameset append open for append noclobber open for writing fails if the directory exists and clobber open for writing recursively deleting the directory if it exits periodicfix Whether to wrap atom positions across periodic boundaries to minimize bond lengths Boolean cen
186. need not evaluate angle_harm records that are marked as constrained These terms are in the angle Hamiltonian category 7 2 3 Proper dihedral terms Desmond implements two functional forms for calculating proper and improper torsion potential terms The first is 6 Vi Giger feo gt fen cos ndija1 do n 1 where feo fes are dihedral angle force constants in units of Energy and y is the equilibrium dihedral angle in radians The angle is formed by the planes p0 p1l p2 and pl1 p2 p3 Terms in dihedral_trig are handled by this potential function Table 7 4 Schema for the dihedral_trig table name type description phi0 FLOAT phase DEGREES fc0 FLOAT order 0 force constant ENERGY fel FLOAT order 1 force constant ENERGY fc2 FLOAT order 2 force constant ENERGY fc3 FLOAT order 3 force constant ENERGY fc4 FLOAT order 4 force constant ENERGY fc5 FLOAT order 5 force constant ENERGY fc6 FLOAT order 6 force constant ENERGY po INTEGER Ist particle pl INTEGER 2nd particle p2 INTEGER 3rd particle p3 INTEGER 4th particle These terms are in the dihedral Hamiltonian category 7 2 4 Improper dihedral terms The second dihedral functional form is this potential function Vi dizer fe Piznt do where f is the dihedral angle force constant in units of Energy radians and y is the equilibrium dihedral angle in radians The an
187. nique For each key key Desmond creates a main loop object of type type The remainder of the table under key contains the object s configuration 18 Chapter 4 Running Desmond Desmond Users Guide Release 3 4 0 0 7 mdsim plugin list my_status my_status type status first 0 interval 1 In this case the mdsim application will create an object of type st atus which is set to run every picosecond Note Main loop plugin objects are evaluated in the order in which they re listed in the configuration In certain cir cumstances listing plugins in a different order can yield different results for example if your simulation calls both the randomize_velocities and eneseq plugins Because randomize_velocities generally changes the ki netic energy of the system different kinetic energies and temperatures are reported if the randomize_velocities plugin is listed before eneseq rather than after the dynamics of the system will be the same but the reported tem peratures will be different Configuring the built in plugins describes the built in main loop plugins 4 3 Running Desmond in parallel Desmond can be run either in serial or in parallel in environments ranging from laptops to large Linux clusters High performance parallel systems consist of nodes connected together in a network containing one or more processors each of which consisting of one or more processor cores or cores In the follo
188. ns The only requirement is that the last statement in a block must be an expression and not an assignment The value of the block is then the value of its last statement Each block introduces a nested scope so that assignments made within a block are only available inside the block and shadow assignments made outside the block An example that uses blocks is the following all pairs interaction s series i 0 length a j O length b r dist a il b 31 Lf oe amp then PAS 6 5 23 k x r2 11 3 Interpreter 101 Desmond Users Guide Release 3 4 0 0 7 else 0 y y Note that the entire body of the enhanced sampling program is treated as if it is wrapped in a block Integer and floating point literals may be used in the normal manner Some functions take strings as arguments This is a special behavior and strings do not exist anywhere else within the interpreter 11 3 2 Interpreter values All values within the interpreter are arrays of germs A germ is a double precision value and its differential The differentials are not manipulated directly by the user instead every function uses the differentials of its arguments to compute the differential of its return value In this way the force associated with the user specified potential is computed automatically Numeric literals in a symbolic expression are converted internally to arrays of length with zero differential Some functions take an integer
189. ntum rescalings to mitigate the perturbation to the dynamics In Desmond the net energy or heat added or subtracted by the antithetic thermostat is accounting for in the extended variable energy term which results in a conserved energy useful for diagnostic purposes Table 9 6 Configuration for Anti NVT name description thermostat bath_dof The number of degrees of freedom in the stochastic baths Optional defaults to 0 Integer gt 0 thermostat seed The random number seed for Gamma distributed variables Optional unless bath_dof is given Integer gt 0 9 6 4 L_NVT Langevin constant volume and temperature The L_NVT dynamical system is configured as shown in integrator L_NVT thermostat tau T seed s It supplies a thermostat using the Langevin method for all of the elements of the integrator temperature list This dynamical system adds a damping term and a stochastic force to the atoms The amount of stochastic force used is a function of the T for the jt thermostat while the damping 1 7 is uniform across all thermostats The mean collision time for water roughly 1 62 ps is often used for 7 The dynamics are given by the SDE Ti p m gt Bi VA U r Bi 05 t 7 where each component of the random vector S t is a standard Wiener process W t having the probability density Prob w lt W t lt w dw exp w 2t dw 1 v2rt and o y 2m kgT T wh
190. o static potential Vias Lert riy V20 2 Both methods compute the sum of far interactions for all pairs of particles including those pairs that are excluded Thus it is necessary to subtract the portion of the energy and forces due to the exclusions with a far exclusion computation The Ewald mesh dimensions are specified as the number of subdivisions k along the axes of the global cell The spacing of mesh points along the axis for example is k A mesh spacing between 0 6 A and 1 5 A usually gives a good balance between accuracy and efficiency The subdivisions are required to be integers of the form ki 2 3 5 74 for nonnegative integers a b c and d that are also multiples of the global cell partition along the corresponding axes see The Global Cell the smallest such integer that provides a suitable mesh spacing is recommended The Fourier c2c and r2c differ in their efficiency depending on the underlying networking hardware Since the type of the network is not available to Desmond the user is responsible for picking the correct value of transform For low latency networks such as available with InfiniBand we have found that setting transform c2c gives the best performance at high levels of parallelism with transform r2c performing better at low levels of paral lelism and transform r2c 2round at very low levels For high latency networking hardware such as Gigabit Ethernet setting transform r2c has been found
191. o requests for temperature changes Plugin setup Within the application specific plugin section the following specifies the component of annealing that schedules the temperature changes app plugin list key key type anneal first to interval 6 schedule time ti to tn value Ti Jo Tn e to first time to reset the thermostat temperature e interval between thermostat resets There is a small performance cost to resetting the thermostat so its recommended that the delta be set no smaller than the natural thermalization time of the system typically on the order of 1 ps e schedule When the plugin is invoked as specified by first and interval a target temperature is com puted based on the current chemical time t If t lt t no action is taken If t gt ty the target temperature will be Ty the last temperature in values Otherwise the target temperature is computed by linearly interpolation between the time points t t 1 that bracket the current time gt tts a Ait Fa aa For example the following schedule would heat a system from 0 to 500 K during the first 20 ps then cool it to 300 K during the subsequent 80 ps and maintain it at 300 K thereafter mdsim plugin keyx schedule time 0 20 100 value 0 500 300 4 6 2 Biasing Force The BiasingForce plugin can be used to restrain two groups of atoms within a chemical system with respect to eac
192. ocess per processor core Each application logs its configuration at startup so users can observe the net result of the configuration options This includes displaying a list of the loaded plugins with full paths so that you can see all the code that Desmond can access Plugins are described in Configuring the built in plugins Table tab clo lists the full set of supported options All command line options have the same effect for all applications except restore which pertains to the mdsim and remd applications only enabling them to start from a check point file It is an error to provide a command line option that is not recognized by Desmond or one of its components Command line options can be given in any order Table 4 1 Desmond command line options argument description tpp N Sets the number of threads per process Defaults to 1 cpcN Gives the number of cores per physical chip Defaults to 0 spin N Sets the worker thread idle strategy Defaults to O If 1 or 2 then use spin lock based idle strategies Sets the name of the communications plugin to use for parallel jobs destrier name Defaults to serial include file name Adds configuration information from the given file Can be given any number of times cfg string Adds configuration information from the given string Can be given any number of times restore file Restarts the mdsim or remd applications from a checkpoint B
193. of additional outer step dynamics Integer gt 0 9 4 Pressure Some dynamical systems change the unit cell vectors of the global cell thus changing the size and possibly the shape of the cell during the integration to realize a constant pressure ensemble The pressure section gives the parameters for such systems Pressure is configured as shown in integrator pressure isotropy isotropic semi_isotropic anisotropic constant_area max_margin_contraction Cmax P_ref Po tension_ref t33 isotropy constrains the changes allowed for the global cell e isotropic The cell scales uniformly along all three axes e semi isotropic The X and Y axes scale uniformly while the Z axis scales independently e anisotropic The cell scales independently along all three axes e constant area The cell scales along its Z axis only As the cell changes shape its clone radius changes as well If the new radius is less than a factor of Cmax times the old radius certain lazily updated quantities such as particle pairlists are immediately recomputed Pp and t33 are parameters that appear in the equations of certain dynamical systems Their roles in those systems are described in Dynamical systems 72 Chapter 9 Dynamics Desmond Users Guide Release 3 4 0 0 7 Table 9 3 Configuration for pressure name description isotropy The allowed class of cell changes Symbol max_margin_contraction The amoun
194. oms cos Class Threaded Arguments a array Return the element wise cosine of a cross Class Normal Arguments e a 3 element array e b 3 element array Return the cross product of a and b delta Class Normal Arguments e gidl particle e gid2 particle Return the minimum image displacement from particle gid1 to particle gid2 dihedral Class Normal Arguments e a 3 element array 146 Chapter 19 Enhanced sampling function reference Desmond Users Guide Release 3 4 0 0 7 e b 3 element array e c 3 element array Return A 2 element array The first element is the cosine of the dihedral angle of vectors a b and c and the second element is the sine of the dihedral angle dihedral_gid Class Normal Arguments e pl particle e p2 particle e p3 particle e p4 particle Return A 2 element array The first element is the cosine of the dihedral angle for particles p1 p2 p3 and p4 and the second element is the sine of the dihedral angle dihedral_gid_radians Class Normal Arguments e pl particle e p2 particle e p3 particle e p4 particle Return Dihedral angle in radians for particle p1 p2 p3 and p4 Result is in the range 7 7 This function is based internally on atan2 Please see the documentation for atan2 for more information In particular this function is discontinuous when the dihedral angle is near 7 and the derivative of the function does not describe this
195. on is evolved according to a dynamical system specified by the integrator type variable which is a name This name selects the system to be used and is also treated as a key in the integrator section under which the parameters for the specified system can be found 9 6 1 V_NVE Verlet constant volume and energy The V_NVE dynamical system is configured as shown in integrator V_NVE No parameters are needed The system is the ODE Ti P mi Pi Vr U r which conserves the scalar Ho r p Y 15 12 2m U r and the phase space density differential form w PRPA 2 where d37 and d p are the volume elements of the position and momentum of particle i Thus the trajectory if ergodic is expected to sample uniformly from a surface of constant Ho r p 9 6 2 NH_NVT Nos Hoover constant volume and temperature The NH_NVT dynamical Mar 1992 system is configured as shown in integrator NH_NVT thermostat mts m tal ma sin tml This system supplies a thermostat using a Nos Hoover chain with extended system variables for each of the elements of the integrator temperature list the length of which must match that of the thermostat list For each thermostat and each 7 parameter a pair of variables 1 is introduced for a total of 2nk additional variables k 74 Chapter 9 Dynamics Desmond Users Guide Release 3 4 0 0 7 being the number of thermostats The dyna
196. on options You add constraints to a structure file using the build_constraints program provided with Desmond Note Viparr does not update the constraint tables in a dms file so if you use Viparr to update a structure file that included constraints you ll need to add the constraints again build_constraints examines a structure file for atom groups of the following forms e AHn An atom other than hydrogen bonded to n hydrogen atoms HOH An oxygen atom bonded to two hydrogen atoms and no other atoms Desmond s implementation of constraints is described in Constraints The atom identities are determined from the atomic number of each atom while the bonds are determined from the bond table build_constraints then constructs a new constraints table replacing any existing table by that name and populates it with the detected constraint groups By default the stretch and angle force terms corresponding to groups that are constrained by the constraint groups are also modified the constrained column of stretch_harm and angle_harm records is set to 1 This is done because evaluating forces on constrained groups is wasted effort the constrained lengths and angles are not allowed to change However the constrained bonds and angles cannot be completely removed from the structure file because the minimize application does not currently evaluate constraint terms but instead evaluates the forces in the constrained bond and angle terms T
197. ons see Naming output files with individual frames written as files within that directory as described in Trajectory Format and Analysis The periodicfix center and glue options together describe how simulation coordinates should be prepro cessed before being written to the frameset If periodicfix is turned off and no centering or glue is applied all atom coordinates are wrapped to the central unit cell irrespective of bonds between atoms This makes visualization and trajectory analysis difficult The periodicfix option instructs Desmond to re wrap atoms so that no bond is longer than half the length of of any global cell vector Note that the bonds considered are those of the bond section of the structure file not the st retch_harm or any other force field terms The glue option extends the list of bonds supplied by the structure file with fictitious bonds that can improve the wrapping behavior For example if a protein is composed of four disconnected monomers that nevertheless stay non covalently bound to each other during a simulation it will be desirable to keep them together during wrapping Without glue however if one monomer strays close to the edge of the periodic cell it will be wrapped to the other side while the other three monomers remain where they are To correct this behavior one could use a glue configuration of the form glue nl n2 n3 n4 where each n is from a different monomer This would create fictitious bonds n1
198. ons exert on a particle varies more slowly in time than other interactions you can configure Desmond to compute them less often to further accelerate the computation this is discussed below in Dynamics 3 3 Particles Desmond represents each atom in the chemical system as a particle Special cases for molecules such as water are discussed below see the discussion of pseudoparticle see pseudoparticles The particle e models key real world aspects of an atom its mass charge position and velocity e participates in bonds of specified types and e can be a member of one or more groups You can assign particles to groups for various purposes e To understand how energy is distributed throughout the system particles can belong to different energy groups e To control the temperature of subsets of particles independently particles can belong to different temperature groups e To restrain them to a predetermined position relative to another particle group or to the simulation coordinate system particles can belong to a center of mass group To hold them motionless in the simulation particles can belong to the frozen group e To define a ligand used in free energy simulations particles can belong to the ligand group 3 4 Force fields A force field is a model of the potential energy of a chemical system It s a set of functions and parameters used to model the potential energy of the system and thereby to calculate the force
199. oot timestamp a date string resolved to the second taken from the start time of one of the parallel processes s A sequence number an integer starting at zero and incrementing each time this filename is expanded producing an ordered sequence of files rather than overwriting the same file P The UNIX process ID of the process writing the file as a hexadecimal integer R The rank a unique identifier within a parallel run of the process writing the file F S The result of passing S to strftime ee The character For example if you wish to write an output file several times during a run the filename my_output sS creates a sequence of files named my_output 0 my_output 1 and so on The last used value of the sequence number is saved in the checkpoint file To ensure that unique files are created with each Desmond run give files names such as my_output QB thus causing each to be named with a unique timestamp If the runs are expected to take less than one second to complete unique file names would require a different strategy perhaps my_output B P A filename can encode the current date and time in various formats For example you can use a file name of the form my_output tF Y m d to name a file according to the current date my_output 2010 04 23 You could name your checkpoint file in this way 1f you wanted to ensure that no more that one checkpoint file is written per day Plugins that periodical
200. ors angle_gid cosine of angle for 3 particles angle_gid_radians angle of 3 particles unstable for angles near 0 or 7 angle_radians angle of 2 vectors unstable for angles near 0 or 77 array create array atan2 arctangent for two arguments center_of_geometry center of geometry for a group of particles center_of_mass center of mass for a group of particles contact_map contact map for a group of atoms cos cosine cross cross product Continued on next page 11 3 Interpreter 103 Desmond Users Guide Release 3 4 0 0 7 Table 11 1 continued from previous page delta min image vector between two particles dihedral cosine and sine of dihedral angle for 3 vectors dihedral_gid cosine and sine of dihedral angle for 4 particles dihedral_gid_radians dihedral angle for 4 particles problematic for angles near 7 dihedral_radians dihedral angle for 3 vectors problematic for angles near 77 dist min image distance between two particles dot dot product exp exponent gibbs_max a softened version of the minimum gibbs_min a softened version of the minimum helix helicity length array length log logarithm mass mass of particle in amu meta metadynamics min_image minimum image of vector mod modulus ncoordination returns the coordination number between two groups of atoms norm norm of vector norm2 squared norm p
201. os lookup particle position pos_inner_prod weighted sum of particles positions pow positive base raise to arbitrary power print create output rad_gyration radius of gyration for a group of atoms rmsd RMS displacement from model structure rmsd_torsion torsional rmsd for a group of atoms sign sign function with sign 0 1 sin sine sqrt square root store store value for use at a later time step sum sum an array time chemical time 11 3 6 Periodic Images The interpreter does not distinguish between vectors representing atom coordinates and arbitrary length 3 arrays and the user is responsible for considering periodic images when computing collective variables In particular the code pos gid 2 pos gid 1 will not compute the minimum image displacement due to wrapping of periodic coordinates The function min_image will compute the minimum image of an arbitrary length 3 array for the current simulation box As a convenience the functions delta and dist cover the most common coordinate differences that are needed They are defined as follows delta gid2 gidl min_image pos gid2 pos gidl dist gid2 gidl norm min_image pos gid2 pos gidl For algorithms that operate on widely separated parts of the protein such as center of mass the user is strongly encouraged to consider carefully how periodic images will be handled Note that functions like the Enhanced Sampling implementation of RMSD have carefu
202. oses of the center of geometry is then the sum of all these minimum image displacements for each adjacent pair in the gid array going back to the first particle The user must guarantee that each adjacent pair in the position_gid array is less than 1 2 of the unit cell apart If this condition is violated then particles may be wrapped to the wrong side of the cell distorting the center of geometry center_of_mass 145 Desmond Users Guide Release 3 4 0 0 7 Class Normal Arguments e gids array of gids Return center of mass with periodic image handling To compute the center of mass with periodic image ambiguities the following convention is used The minimum image displacement of each GID is calculated with respect to the previous GID in the gids array The location of a particle for the purposes of the center of mass is then the sum of all these minimum image displacements for each adjacent pair in the gid array going back to the first particle The user must guarantee that each adjacent pair in the position_gid array is less than 1 2 of the unit cell apart If this condition is violated then particles may be wrapped to the wrong side of the cell distorting the center of mass contact_map Class Normal Arguments e r distance threshold e gids list of particles Return The number of residues in the list gids within distance r of each other This function counts up the number of elements in a contact map within a groups of at
203. ot have any nonbonded interactions name should have the special value none The parameters for nonbonded interactions will be stored in a table called nonbonded_param whose schema depends on the value of name in nonbonded_info All such schemas must have a primary key column called id there are no other restrictions The nbt ype column in the particle table gives the nonbonded type assignment The value of the type assignment must correspond to one of the primary keys in the nonbonded_param table Typically the parameters governing the nonbonded interaction between a pair of particles is a simple function of the parameters assigned to the individual particles For example in a Lennard Jones functional form with parameters sigma and epsilon the combined parameters are typically the arithmetic or geometric mean of sigma and epsilon The required approach is obtained by the application from the value of rule in nonbonded_info For the interaction parameters that cannot be so simply derived a table called nonbonded_combined_param may be provided with a schema shown in Schema for the nonbonded_combined_param table Like the nonbonded_paranm table the schema of nonbonded_combined_param will depend on the functional form of the nonbonded interactions but there are two required columns which indicate which entry in nonbonded_param are being overridden 130 Chapter 17 Appendix DMS file format Desmond Users Guide Release 3 4 0 0 7 Ta
204. oup A Group A consists of all atoms whose grp_bias property is set to the integer value A Group B those set to the integer value B see Preparing a structure file The allowed values of grp_bias are0 1 2 and 3 and by default all particles in the chemical system are in center of mass group 0 If A 1 however BiasingForce will restrain group B with respect to the simulation box 4 6 Configuring the built in plugins 27 Desmond Users Guide Release 3 4 0 0 7 The BiasingForce plugin works by imposing the following restraining potential on the molecular system Ra gt K 5 gt AS gt 5 Ran Gai 2 y Ras das y gt Ras das 2 4 2 Ebiasing k 5 Raz Ry Ww 7 _ w _ 5 w 7 E S Gae1 da 1 5 G p2 das 1 T Gaps das 1 where Ra Rp is the center of mass of group A B Rap 14B YAB 24B Rp Ra Aa A 1 2 3 GBa are the principal axis of group A B and G is the rotational matrix that will superimpose the qg s onto q4 s when B is in the relative orientation with respect to A as specified by the Euler angles 0 w In Ebiasing the first three terms restrain the relative displacement between the centers of mass of groups B and A to the desired displacement x y z The term in the second line restrains the scalar center of mass distance R4 pg of the two groups A and B and R is the target displacement The three terms in the third line re
205. pattern These APIs take the form of abstract C classes which are subclassed to create the new functionality and extensible factories that can construct instances of these classes When an extension is loaded the plugins in the extension add new concrete types in the extension to various factories in Desmond The most common factories are listed in common Desmond factories 109 Desmond Users Guide Release 3 4 0 0 7 Table 12 1 common Desmond factories factory in namespace Desmond description MainPlugin factory main loop plugins Integrator factory integration algorithms Hamiltonian factory force terms App abstract_driver factory Application type e g mdsim remd 12 2 Running your plugin If your plugin resides in a separately compiled extension Desmond must find it and load it before it can be used When Desmond starts it searches for extensions by parsing the environment variable DESMOND_PLUGIN_PATH and searching for shared libraries created according to the compilation guidelines outlined in Implementation Extensions are loading immediately after the Desmond executable starts Desmond processes extensions in three steps 1 Desmond reads the extension s type description boot and halt methods This information is created by a static instance of the desres plugin declaration class When Desmond loads the extension it examines this information and checks to see if
206. pecifies the application and the structure file as in app mdsim boot file path to my input dms The cfg option allows you to append additional configuration information to the command line It s often used to specify the structure file For example to invoke desmond with the structure file path to my input dms desmond includ quil cfg cfg boot file path to my input dms This has the same effect as the line from the configuration file above Note Use quotation marks around the parameter to cfg if it contains any special characters such as spaces that are interpreted by the shell You can specify multiple configuration files this can be useful for configuring a simulation in a modular way For example you might choose to have alternative integrator configurations in two files named nve cfg and ber_nvt cfg with other configuration parameters in the base configuration file in base c g Then for a simu lation in which you intend to use the Verlet constant volume and energy integrator you d invoke desmond include base cfg include nve cfg cfg boot file input dms Whereas for a simulation in which you intended to use the Berendsen constant volume and temperature integrator the command line would instead be desmond include base cfg include ber_nvt cfg cfg boot file input dms You cannot specify multiple structure files The include and cfg arguments are evaluated in order and t
207. pports specification of an interaction between groups of particles where the force acting on the particles is based on the distance between the geometric centers of the particles in the respective groups A given particle can participate in multiple groups The flat bottomed harmonic stretch term also differs from the other potentials in that it transitions from a harmonic to a linear potential at large separation The harmonic stretch term is designed to model ambiguous NOE distance restraints The treatment of ambiguous NOE data was discussed by Nilges in his JMB _245_ p 645 1995 and is also presented in Nilges s contribution chapter 13 in the book Computational Biochemistry and Biophysics edited by Becker et al 2001 In an ambiguous distance restraint the measured NOE distance is assigned to a r REXP summed distance with expo nent REXP 6 In what follows we leave REXP flexible although its meaning is not well defined for values other than 6 The restraint energy calculation has the following form d 0 for all i in A and all j in B xij min_image xi xj yij min_image yi yj zij min_image zi zj rij sqrt x1ij 2 yijy 2 zij 2 d rij REXP end for d d 1 REXP 7 2 Bonded pair and excluded interactions 55 Desmond Users Guide Release 3 4 0 0 7 The energy is then computed as LE dy d lt L B ix 0 L lt d lt U AA TE U U lt d lt U s a b d U c d U U s lt d wh
208. r Reut can have detrimental effects on the simulation For those potential functions that are computed by a piecewise polynomial interpolation function for the default near term this is only the electrostatic potential you can alleviate this detrimental effect somewhat by choosing a tapering strategy where the potential being approximated by d is first replaced with a function r and constructed to approximate dy instead Three strategies are available for constructing r shift or r r Reut which vertically shifts the function so that r r 0 for r Reut cl switch and c2switch r r 1 2 20 0 1 r r 1 a 1 3x 62 6 r respectively for Rtap lt r lt Reut Where x r Rtap Reut Rtap and pr r p r forr lt Rrap In practice tapering is not usually necessary for typical cutoff values Reut 10A is typical Piecewise polynomial interpolation is used as an approximation for some potentials in the near term which ones dependent on the kind of near term chosen The interpolation is actually a piecewise cubic spline interpolation db of constructed as function of r interpolating through points of the form r R2 am nz where 0 lt m lt nz As such the accuracy of the approximation is controlled through n set by the parameter n_zone Although a simple error bound is difficult to express for general for power law potentials x r for some C the relative error of
209. r The interaction between A and B is computed on the process containing their midpoint 40 Chapter 5 The Global Cell Desmond Users Guide Release 3 4 0 0 7 In the example illustrated in the 2D illustration either process could in principle compute the AB interaction In Desmond the process that does so is the one whose home box contains the midpoint between the two particles After computing forces on the clone it sends the result to process B which sums A s result with its own before computing B s new position and velocity More generally the process that computes an interaction of a group particles is the one whose home box contains the unweighted midpoint of that group At the end of a timestep after new particle positions are computed some particles will have moved out of their previous home boxes into neighboring ones Migration is the process by which particles are reassigned to the processes responsible for their new home boxes You can configure Desmond to migrate particles every time it updates particle positions at each inner timestep a setting called eager migration However during migration processes need to exchange a lot of data so it s desirable to minimize its occurrence Lazy migration lets you avoid some communication overhead by reassigning particles to home boxes less often than every time particle positions are updated Position updates can then occur without migration The migration schedul
210. rbed_ct that is that disappear in the perturbed_ct i_fepio_aj is set to 1 For atoms in the perturbed_ct that map onto dummy atoms in the original_ct we assign atom numbers i_fepio_ai counting from the number of real atoms in the original_ct 1 For instance if ten real atoms are in the original_ct these dummy atoms are numbered i_fepio_ai 11 12 and so on fepio_bondmaps This indexed block maps the bond potentials from the original_ct onto the perturbed_ct 138 Chapter 18 Legacy Applications Preparing a Maestro structure file Desmond Users Guide Release 3 4 0 0 7 Table 18 7 fepio_bondmaps properties property name description 1 fepio_ti Bond potential number in original_ct Negative bond number indicates a bond involving at least one dummy atom 1 fepio_tj Bond potential number in perturbed_ct Negative bond number indicates a bond involving at least one dummy atom 1 fepio_ai The first atom in the bond in original_ct Negative atom numbers can appear here by the same convention as in atommaps i_fepio_ai The second atom in the bond in orignal_ct Negative atom numbers can appear here by the same convention as in atommaps fepio_anglemaps This indexed block maps the angle potential from original_ct onto perturbed_ct Table 18 8 fepio_anglemaps properties property name description i_fepio_ti Angle potential number in original_ct Negative angle numb
211. resent between a pseudoatom and its parent particle or particles these bonds aid in visualization The p0 and p1 values correspond to an id in the particle table Each p0 p1 pair should be unique non NULL and satisfy p0 lt pl 17 1 3 The global cell Table 17 4 Schema for the global_cell table name type description id INTEGER vector index 0 1 or 2 x FLOAT x component in LENGTH y FLOAT y component in LENGTH Z FLOAT z component in LENGTH The global_cell table specified in Schema for the global_cell table specifies the dimensions of the periodic cell in which particles interact There shall be three records with id 0 1 or 2 the primary key is provided since the order of the records matters and one would otherwise have difficulty referring to or updating a particular record in the table 17 1 4 Additional particle properties Additional per particle properties not already specified in the particle table should be added to the particle table as columns Optional particle properties shows the schema for the additional properties expected and or recognized by Desmond and by Viparr 128 Chapter 17 Appendix DMS file format Desmond Users Guide Release 3 4 0 0 7 Table 17 5 Optional particle properties name type description mass FLOAT Desmond particle mass in MASS charge FLOAT Desmond particle charge in CHARGE vx FLOAT Desmon
212. royalty free license on a non exclusive non transferable non assignable and non sublicensable basis to install and use for non commercial research as defined below the molecular dynamics software program known as Desmond Version 3 including any version of such program whose version number begins with 3 and any associated documentation any such documentation and any such version collectively referred to herein as the SOFTWARE The SOFTWARE may be accessed held or otherwise used only with a valid license and this Agreement confers a valid license only to a academic or other not for profit research entities and b individuals who are affiliated with such entities in each case a and b provided that such entities and or individuals use the SOFTWARE exclusively for non commercial research purposes Upon any change in LICENSEE s status as or affiliation with a not for profit research organization or in LICENSEE s use of the SOFTWARE exclusively for non commercial research all licenses granted hereunder shall terminate immediately with or without any notice by DESRES If LICENSEE wishes to continue using the SOFTWARE after any such termination LICENSEE must apply for a new SOFTWARE license any approval of which application shall be at DESRES sole discretion Use of the SOFTWARE is restricted to non commercial research conducted by LICENSEE and if LICENSEE is an organization LICENSEE s employees research advisees and student
213. rray The location of a particle for the purposes of RMSD is then the sum of all these minimum image displacements for each adjacent pair in the position_gid array going back to the first particle The user must guarantee that each adjacent pair in the position_gid array is less than 1 2 of the unit cell apart If this condition is violated then particles may be wrapped to the wrong side of the cell distorting the RMSD Note that the derivatives of the model configuration and the weights are not considered in computing the derivative of the RMSD Model coordinates are not wrapped in any way rmsd_torsion Class Normal Arguments e gids list of particles 4x length of phis e phis list of dihedral angles in radians Return X i 6 N 1 2 where Y is the dihedral angle of the group of 4 atoms and N is the length of the list of ds The rms average difference of the dihedral angles of the atoms taken 4 at a a the gids list and the angles from the phis list sign Class Threaded Arguments e a array Return For each element x of a returns 1 if x gt 0 1 if x lt 0 and x otherwise Sign is computed element wise sin Class Threaded Arguments a array Return the element wise sine of a sqrt Class Threaded Arguments a array Return the element wise square root of a store Class Special Form Arguments e storename see description e a array 152 Chapter 19 Enhanced sampling func
214. s Authorized Users The term non commercial research means any academic or other research which x is not undertaken for profit and y is not intended to produce results works services or data for commercial use by anyone Any other parties including without limitation any collaborators of LICENSEE wishing to install or use the SOFTWARE may do so only if such parties have executed a separate license agreement with DESRES giving such parties the right to do so DESRES reserves all rights not expressly granted herein T T 155 Desmond Users Guide Release 3 4 0 0 7 L IC ENSE hereby represents and warrants 2 Representations and Warranties that a LICENSEE has the necessary aut all information that LICENSEE connection with LICENS qualifies for t E basis of the criteria sp conditions set forth in 34 LIC Restrictions ENS may make copies of t and d hori has provided or will hereafter provide in this Agreement is and will be correct and complete he non commercial licens cified herein will ensure that all of its Authorized Users abide by his Agreement for bona fide backup or arc
215. s If an exchange is accepted the velocities of the replicas are rescaled to the temperature of the host replica otherwise the positions are simply swapped back Thus in any replica the temperature and Hamiltonian will stay the same but the dynamics will be discontinuous as new coordinates are swapped in via exchanges The cfg configuration in remd serves two purposes First the number of entries in the list r serves to specify how many replicas are to be run in the simulation Second each entry in remd cfg overrides the configuration for the corresponding replica in the same way that the cfg command line option overrides a setting for an mdsim run For example cfg integrator temperature T_ref 300 plugin eneseq name 0 ene integrator temperature T_ref 303 3333 plugin eneseq name 1 ene integrator temperature T_ref 306 6667 plugin eneseq name 2 ene integrator temperature T_ref 310 plugin eneseq name 3 ene has four replicas replica 0 will see a configuration with the integrator temperature set to 300 replica 1 will get a temperature of 303 3333 and so forth Also in this example a plugin variable is overridden on a per replica basis Overrides to the remd section itself should not qualified with the prefix remd as one would have expected 4 4 3 remd graph The remd graph app driver is a generalization of the remd driver intended to give advanced users more control over the set of possible exchanges
216. s Same length as schedule time list List of Scalars 4 6 4 energy_groups Periodically writes energy to the output file p broken down both by the energy group property of the particles and the Hamiltonian category property of the potential energy term Energy groups are assigned in the structure file while the string identifiers of various Hamiltonian categories are set by their computational pipelines Additionally the 3 x 3 instantaneous pressure tensor and the nonbonded correction energy are printed if specified The nonbonded correction energy is the sum of the nonbonded tail correction and the electrostatic self energy correction see Nonbonded tail corrections and Electrostatic self energy correction app plugin list key key type energy_groups first ty interval ti name p options pressure_tensor corr_energy self_energy The output format is a sequence of ascii blocks of plain text Each block begins with a line of the form time t en e E_p p E_k E_x P P V V giving the chemical time the raw potential the potential kinetic and extended energies as well as the pressure and volume The raw potential energy is potential energy without the electrostatic self energy or the nonbonded tail corrections What follows is then a break down of the raw potential energy by energy group The kinetic energy is broken down into the kinetic energy per gro
217. s frames can include system charac teristics such as its total energy temperature volume pressure and dimensions of the global cell You can configure Desmond to output frames typically at an interval corresponding to a multiple of the outer timestep when nonbonded far interactions are computed A time ordered series of frame files representing the dynamics of the chemical system for the specified time period Framesets are ordinarily the meaningful unit of analysis for vrun or other analysis applications such as VMD 12 Chapter 3 Key Concepts Desmond Users Guide Release 3 4 0 0 7 3 8 4 Workflow The following typical workflow illustrates the roles of Desmond s three main applications as well as those of other cooperating applications 1 Prepare the structure file Typically start with a Protein Data Base pdb file and produce a DMS file a Depending on its contents and the manner in which it was created it may need some repair of artifacts e g due to x ray crystallography Maestro is one tool that can do this others also exist Maestro or a comparable application outputs a structure file typically containing e the solute proteins ligands or other molecules of interest e the solvent water and often ions such as sodium potassium or chloride to ensure that the overall chemical system is neutral with respect to charge A charge neutral system is desirable for computing long range electrostatic interact
218. s given by piecewise linear interpolation as in Equation 4 3 For example if a schedule has time points 1 10 and values 1 0 0 0 then the scale factor will be 1 0 for times before ps 0 0 for times after 10 ps and in between the scale factor will decrease linearly 4 6 9 pprofile The pprofile plugin computes pressure profiles which gives the surface tension in a molecular system as a function of the z coordinate Pressure profile analysis can give insight into the role of the lipid environment on embedded proteins Configuration of the pprofile plugin is shown in the Example app plugin list key key type pprofile first tf interval ti eval_interval te nslabs y name p include optional exclude optional 4 6 Configuring the built in plugins 33 Desmond Users Guide Release 3 4 0 0 7 Table 4 17 Configuration for pprofile name description first first output time Time interval interval between outputs Time eval_interval time interval between virial calculations Time nslabs number of simulation cell partitions Integer name frameset output directory String include 1f present include only the given force terms in the virial calculation List of strings exclude if present do not include the given force terms in the virial calculation it is an error to specify both include and exclude in the same pprofile instance L
219. s of command line options or configuration file parameters according to a uniform syntax The command line options can be summarized and stored in plain text files called configuration files which represent a summary of the state of the configuration This is discussed in Invoking Desmond This appendix describes configuration file syntax in formal terms and provides additional examples In Backus Naur Form BNF a configuration is CONFIG gt KEYVAL KEYVAL gt key VALUE VALUE gt CONFIG VALUEx x atom The terminals are e key A valid key name a string consisting of alphanumerical and underscore characters with a nonnumerical leading character e atom An arbitrary string e A nil value Not commonly used The keys of a CONFIG are assumed to be distinct and its KEYVAL terms are considered unordered A configuration is therefore a table of atoms lists and more tables Because of this tree structure leafs and subtrees can be referenced by a path starting from the root KEYPATH gt key INDEX INDEX gt number key A key indexes a table A subscript number accesses a zero based list This is identical to the member vector indexing notation used in many programming languages A subscript expression of the form can be used in assignments A keypath is a path to a configuration parameter For example force nonbonded far sigma is a keypath referring to the si
220. s on each particle To accurately simulate different kinds of systems Desmond supports several variants of the Amber CHARMM and OPLS AA force field models see Force fields built into Viparr To more accurately simulate the behavior of water or other molecules certain force fields add electrostatic or van der Waals interaction sites located where no atom is Desmond implements these as pseudoparticles Desmond supports the most common kinds of pseudoparticles including those needed for common water models such as SPC TIP3P TIP4P and TIPSP See details in Virtual sites Like particles pseudoparticles have a mass charge position and velocity however their mass is often zero 3 3 Particles 9 Desmond Users Guide Release 3 4 0 0 7 3 5 Space The volume of space in which the simulation takes place is called the global cell A three dimensional volume of space containing the chemical system This volume is ordinarily visualized as a three dimensional rectangular box though Desmond can simulate other shapes The simulation can change dimensions in the course of running for example to satisfy a requirement for a constant pressure Figure 3 2 The global cell is a three dimensional parallelpiped with periodic boundary conditions Positions within the global cell are specified in x y z coordinates Desmond employs a technique known as periodic boundary conditions to wrap each face of the global cell to its opposite face T
221. s used to store dimensionful quantities we must declare a system of units The units in DMS summa rized in DMS system of units reflects a compromise between an ideal of full consistency and the reality of practical usage in particular the mass unit is amu rather than an algebraic combination of the energy length and time units Table 17 1 DMS system of units TIME picosecond CHARGE electron charge LENGTH Angstr m A ENERGY thermochemical kcal mol MASS atomic mass unit amu In addition to tables DMS files may contain stored queries known as views A view combines data from one or more tables and may apply a predicate as well a sorting criterion How this is actually done in SQL will be obvious to database afficiandos for this specification it suffices to note that a view looks just like a table when reading a DMS file so the views will be described in terms of the data in their columns just as for tables Importantly views cannot be written to directly one instead updates the tables to which they refer 17 1 Molecules The DMS file contains the identity of all particles in the structure as well as their positions and velocities in a global coordinate system The particle list includes both physical atoms as well as pseudoparticles such as virtual sites and drude particles The most important table has the name particle all other tables containing additional particle properties or particle particl
222. served quantity but in this case the conserved quantity is enthalpy To describe the equations of motion in Piston_NPH recall the definition B a b cj a 3 x 3 matrix with the system s unit cell vectors as columns the volume of the system is the determinant B Since changes in the global cell affect long range interactions we expose the dependence of the potential function on B writing U r B for the potential energy and writing Ho r B p gt pi 2m U r B The dynamics of the cell are expressed through some number of new scaling variables s sg and their relative momenta n Na depending on the pressure isotropy For a given isotropy we define the maps B A and a a is the adjoint of A as shown in these equations Isotropic S1 0 0 a B s1 0 S1 0 a b c 9 3 0 0 1 m 0 0 A m 0 m 0 0 0 m ma M2 Maiz aj ma Ma Ma3 M11 M22 M33 31 M32 M33 78 Chapter 9 Dynamics Desmond Users Guide Release 3 4 0 0 7 Semi isotropic S1 0 0 4 B si s2 0 si 0 a b 9 4 0 0 S2 m O0 0 A m n2 0 m 0 O One Mii Mi2 M13 E mii mae a mar M22 M23 M33 m31 M32 M33 Anisotropic S1 0 0 B s1 2 53 0 S2 0 Id b e 9 5 0 0 83 m 0 0 A m 72 73 0 m 0 0 On3 Mi 12 Mi3 m11 a M21 Ma Maz M22 M31 M32 M33 M33 Constant area 1 0 S B s 0 1 0 a b d 9 6 0 0 S1 0 00 A m 0 0 0 0 0 m m1 M2 Mis a Ma M22 M23 M33 M31 M32 M33 The ODE for the Pis
223. singularity This can cause significant energy drift when the dihedral angle crosses 7 Users are advised to exercise caution when using dihedral_gid_radians It is preferable to use the function dihedral_gid when possible dihedral_radians Class Normal Arguments e a 3 element array e b 3 element array e c 3 element array Return Angle in radians for vectors a b and c Result is in the range 7 7 This function is based internally on atan2 Please see the documentation for atan2 for more information In particular this function is discontinuous when the dihedral angle is near 7 and the derivative of the function does not describe this singularity This can cause significant energy drift when the dihedral angle crosses 7 Users are advised to exercise caution when using dihedral_radians It is preferable to use the function dihedral when possible dist Class Normal Arguments 147 Desmond Users Guide Release 3 4 0 0 7 e gidl particle e gid2 particle Return the minimum image distance from particle gid1 to particle gid2 dot Class Normal Arguments e a array e b array of the same length as a Return the dot product of a with b exp Class Threaded Arguments e a array Return the element wise exponent of a gibbs_max Class Normal Arguments e T scaling temperature e a array Return T log gt gt exp a T A softened maximum is infinitely differentiable un
224. stone type L_NVT time 100 name secondone type V_NVE time 500 Jee firstone thermostat tau T seed s secondone A typical application would be to have one integrator function as an equilibration or initialization of the second integrator Table 9 16 Configuration for Concatenator name description sequence i type The type of integrator 2 in the sequence e g V_NVE MTK_NPT etc string sequence i time Length of time for which to run integrator 7 in the sequence time sequence i name Arbitrary key name key for the integrator specific configuration information for integrator 7 in the sequence string key The integrator specific configuration section for the particular integrator type of integrator 1 in the sequence configuration 9 6 Dynamical systems 91 Desmond Users Guide Release 3 4 0 0 7 92 Chapter 9 Dynamics CHAPTER TEN FREE ENERGY SIMULATIONS This chapter explains the concepts necessary to configure ligand binding and alchemical free energy simulations and those using the Bennett acceptance ratio method as well as describing how to prepare a structure file for free energy simulations 10 1 Configuring free energy simulations Free energy simulations are configured as shown in force term list key key type alchemical binding alpha_vdw a window ty output
225. strain the relative orientations of the particles in group B with respect to those in group A If the parameters pull_displacement are set to 0 the desired displacement z y z in Equation 4 2 are taken to be 29 Yo 20 in the configuration and they will not change in the course of the simulation But if they are not zero the biasing force will be used to pull the two groups from the initial positions at the specified rates v Uz Uy Vz and the target displacement R t at time t is given by R t a t y t 2 t Ran 0 0t where Rag 0 is the initial displacement between A and B at the beginning of the simulation The same convention applies to pul1_distance and pull_Euler The parameter use_lab_frame_for_displacement is false by default If it is set to true the displacement between groups B and A will be measured in the reference frame of the simulation box and will not be projected onto the reference frame formed by the principal axes of group A Namely the three terms in the first line of Equation 4 2 are replaced in this case by Ke 2 Kz 2 Ky a zap 2 cap 2 yan y A maximum of 4 cm_moi groups can be defined for a chemical system Because the center of mass and moment of inertia are computed for these groups of atoms in order to apply the biasing potential the user must avoid imposing the biasing potential upon a group of atoms that can wrap around the periodic
226. system is configured as shown in integrator Anti_NVT thermostat bath_dof M optional seed s optional 9 6 Dynamical systems 75 Desmond Users Guide Release 3 4 0 0 7 This system supplies the antithetic thermostat a thermostatting dynamics to be described in a forthcoming manuscript It supplements the underlying Verlet dynamics described in V_NVE Verlet constant volume and energy with a discrete thermostatting scheme The kinetic energy K of a system with N degrees of freedom sampling the canonical ensemble with temperature T is Gamma distributed N KaT e k ar Recalling from Temperature that i denotes the thermostat which governs particle i the kinetic energy of the parti cles in the j temperature group is 115 Ri as 2m ilx i j The antithetic thermostat couples each temperature group with a stochastic thermal bath by operating on their com bined energy where the bath energy B is a random variable sampled directly from the gamma distribution N Bj T ket Ny is the number of degrees of freedom of the bath specified by the user If it is not specified it defaults to zero resulting in B 0 and a deterministic algorithm Let F and ERA denote the cumulative distribution function c d f and quantile function respectively of the Gamma distribution Ni N T EL en where N is the number of degrees of freedom of the governed particles in temperature group j Recall from ele m
227. t output dtr w for frame in fs print AT TIME frame CHEMICALTIME 114 Chapter 13 Trajectory Format and Analysis Desmond Users Guide Release 3 4 0 0 7 for attr in frame This iterates over the labels if attr endswith ENERGY print attr getattr frame attr Table 13 1 FrameSet attributes and methods attribute or method description name file name used to open this frameset size number of frames in a frameset also len fs index get index th frame hierarchicalName filename DeepDir hierarchical name of filename framefile frameindex path to filename holding the frameindex th frame frameinfo frameindex framefile filesize time offset framesize of frameindex th frame framefi le is the file of size filesize bytes holding the frame at time The framesize serialized bytes for this frame are at position offset within the file fileinfo frameindex filepath offset framesize first lastpl filesize of the frameindex th frame filepath is the full path name to the file that contains frame index offset is the starting bytes of the frame in the file framesize is the size in case there are varible length frames in a file first is the lowest frame number contained in the same file lastp1 is 1 plus the highest frame number contained in the file 1astp1 itself is not included in the file filesize is
228. t thus effectively continuing a simulation you can specify a reference time equal to the time at which the previous simulation finished Starting with the initial chemical system Desmond 1 computes forces on each particle based on all the other particles in the system and 2 moves the particles according to the results of these computations This sequence forming the basis of the timestep is repeated again and again The period of simulated time computed between each update of the particle positions The action of the force field on the atoms is a continuous function of position and time which the simulation samples at regular intervals Thus the timestep is analogous to the resolution of an image in pixels or the sampling rate of an analog to digital converter And like those it presents trade offs too long a timestep sacrifices accuracy too short performance 10 Chapter 3 Key Concepts Desmond Users Guide Release 3 4 0 0 7 For accurate results the timestep must be short enough to resolve the highest frequency vibrations present in your system sufficiently for the timestepping scheme you are using For typical Desmond simulations timesteps around 1 to 2 femtoseconds fs are sufficient To allow larger timesteps in common situations Desmond also provides constraints discussed in Dynamics 3 7 Dynamics The action of the force field on the particles is described by a differential equation that Desmond integrates nu
229. t as linear as this rationale suggests see below In terms of an overall energy drift more constraint errors manifest as a negative downward drift in conserved energy In fact should one see a substantial negative energy drift one should suspect constraint accuracy In order to guide users away from such problems we have made a table of the energy drifts in Kelvin ns which result from the simulation of a cubic cell of water at standard density for various local cell sizes and inner timesteps For each size and step ten NVE simulations were run with random initial velocities drawn from a Maxwell Boltzmann distribution at 300 K All force interactions were shut off and the constraint convergence tolerances were set very stringently twelve M SHAKE iterations always and thus the simulation is purely that of free motion of rigid wa ter molecules the only possible source of energy being the resolution errors from the constraint calculations The simulations were run for 25 ps of simulated time With the first 5 ps discarded the total kinetic energy profile of each simulation was fit to a line and the drift reported is the mean slope of the ten simulations standard deviation in parentheses For larger time steps and smaller box sizes the simulation energy profiles resembled unbiased random walks and fit poorly to lines as indicated by standard deviations which are larger than their means in these regimes Although real simulations will inc
230. t atom j Pseudoparticle p0 is placed at the fractional position c1 along the interpolated line between p1 and p2 Ic3 virtual site The Ic3 virtual site is placed some fraction a and b along the vectors between particles i j and i k respectively The virtual particle lies in the plane formed by i j k gt Py 1 Ep Ca Cr Corp 7 4 Nonbonded far interactions 63 Desmond Users Guide Release 3 4 0 0 7 Table 7 23 Schema for the virtual_lc3 table name type description cl FLOAT coefficient 1 c2 FLOAT coefficient 2 po INTEGER pseudoparticle id pl INTEGER parent atom i p2 INTEGER parent atom j p3 INTEGER parent atom k fdat3 virtual site The fdat3 virtual site is placed at a fixed distance d from particle i at a fixed angle 8 defined by particles v i j and at a fixed torsion defined by particles v i j k Py Ti tar bro cro x Ti where 7 and 7 are unit vectors defined by Pr T2 X Fk Fj Fe Tj PF The coefficients a b and c above are defined as a d cos 0 b dsin 0 cos p and c dsin 0 sin Table 7 24 Schema for the virtual_fdat3 ta ble name type description cl FLOAT d coefficient c2 FLOAT 0 coefficient c3 FLOAT coefficient po INTEGER pseudoparticle id pl INTEGER parent atom i p2 INTEGER parent atom j p3 INTEGER parent atom k ou
231. t involving the number and masses of all particles based on the Maxwell Boltzmann distribution hence A gt 50t where t is the interval between migrations is sufficient to ensure correct calculations Because Riazy gt Reut additional work roughly of order O R 2 A is needed to iterate over uninteresting pairs for near interactions so for good performance you must strike a balance A typical value used is A 0 625A with a pairlist update every 12 fs though this can miss pairs occasionally The pairlist is updated at each migration event The cutoff radius is a therefore key factor in setting the correct lazy radius and the lazy radius in turn is a key factor in setting the clone radius in particular For correct pairlist assembly Rjazy lt Relone To determine which process computes an interaction between two particles Desmond uses the midpoint method it s the process whose home box contains the midpoint between the two If the midpoint of a pair of particles within the lazy radius lies in a particular home box then in order for both particles owned and cloned to be accessible to the appropriate process the clone radius must be at least half of the lazy radius While the clone radius is set as part of configuring the global cell the cutoff and lazy radii are specified in the force section of the configuration for details see Calculating Force and Energy Note When migrating eagerly t 0 one can set A 0 Reut
232. t migrations Time gt 0 44 Chapter 5 The Global Cell CHAPTER SIX PREPARING A STRUCTURE FILE Starting with version 2 4 Desmond switched the format of its structure file from Maestro to a new format called DMS The DESRES Molecular Structure DMS file format is a set of schemas for storing coordinate and forcefield information about a single biomolecular system in an SQLite format database SQLite sqlite is a in process library that implements a self contained serverless zero configuration transactional SQL database engine The code for SQLite is in the public domain and is thus free for use for any purpose commercial or private Note Legacy Applications Preparing a Maestro structure file provides additional information on the format and contents of MAE files Note As of 3 3 3 0 Desmond can read the old MAE format files An optional configuration parameter boot t ype which defaults to dms can be set to mae which directs Desmond to read the boot file in MAE format SQLite reads and writes directly to ordinary disk files A complete SQLite database with multiple tables indices trig gers and views is contained in a single disk file The database file format is cross platform it is portable between 32 and 64 bit systems between big and little endian architectures and between Unix and Windows operating systems All data in a DMS file lives in a flat list of two dimensional tables Each table has a unique n
233. t of relative contraction beyond which all particle ownerships must be recalculated Real P_ref The reference pressure for the cell Pressure gt 0 tension_ref The reference tension for the cell Optional by default 0 Pressure Length gt 0 9 5 Temperature Each particle in a structure file is assigned a property called its temperature group The following synopsis shows how to assign reference temperatures to sets of temperature groups integrator temperature T_ref T T_groups T_ref Ti groups g we f T_ref Tk groups gr The reference temperature T is taken as the temperature of any component in the system which does not have some other temperature assignment that overrides it For nearly all uses this is the only variable that needs to be set However for certain exceptional applications it is possible to assign alternative temperatures to system components what this means physically is the province of the user It is also sometimes desirable in systems sampling from contant temperature ensembles to assign separate thermostats or no thermostat to subsets of the particles The T_groups section is an optional means for exercising this fine control The elements of the T_groups list correspond to logically distinct thermostats that apply to the temperature groups listed in groups and these groups can be assigned their own reference temperatures 7 Subsequent se
234. t3 virtual site The out3 virtual site can be placed out of the plane of three particles i 7 k gt gt To Ti Cr Fj Ti cal Tk Ti eslrj Ti x Fk Fi Table 7 25 Schema for the virtual_out3 ta ble name type description cl FLOAT coefficient 1 c2 FLOAT coefficient 2 c3 FLOAT coefficient 3 po INTEGER pseudoparticle id pl INTEGER parent atom i p2 INTEGER parent atom j p3 INTEGER parent atom k 64 Chapter 7 Calculating Force and Energy CHAPTER EIGHT CONSTRAINTS This chapter describes the constraints available to eliminate the fastest vibrational motions and how to configure them By applying constraints that eliminate the fastest vibrational motions simulations can be run using longer timesteps typically 2 or 2 5 fs instead of 1 fs Constraints are configured as shown in force constraint tol 06 maxit m use_reshake bp use_Reich br exclude optional include optional Constraints fix the distances between pairs of particles according to a topology of rigid rods Ilr rj diy lr ril det The topologies that can be constrained are e AHn n particles connected to a single particle with 1 lt n lt 8 HOH three mutually connected particles The schemas in the DMS file for AHn and HOH constraints are shown in Schema for the constraint_ahN tables and Schema for the constraint_ho
235. temperature dynamical system Fel 1995 is a combination of the L_NVT see L_NVT Langevin constant volume and temperature stochastic dynamics and Piston _NPH see Piston_NPH constant pressure and enthalpy An additional stochastic differential equation governs the barostat degrees of freedom The SDE for this system is p m A n ri W i nisi W Bi Vz U r B s 1 1 Ng A n pi W Pi 05 t 7 yaa ec p B s Po B s B s oA Dam n aS t 7 where each of the components of the vectors S and g is a standard Wiener process and op y 2W kpg TpTo Although this SDE does not have a conserved scalar it does have an invariant phase space density given by Q f r S P nQ 82 Chapter 9 Dynamics Desmond Users Guide Release 3 4 0 0 7 where f satisfies the PDE 0 D a A iyri W Vz U r 0 1 N9 AM P W Vi f ty va Dif 57Vi 1 7 a ce s B s Po B s B s x gt aati Vif So ms WV 84 Vn nf 30 RV If 7 Tk Tp T then f exp Hp r s p 1 kBT is the invariant phase space density As with L_NVT the energy or heat added or subtracted by the stochastic portions of the SDE are accounted for in the extended variable energy term resulting in a conserved energy useful for diagnostic purposes Table 9 10 Configuration for L_NPT name description barostat tau Used to set the mass see Piston_NPH Time gt 0 barostat T_re
236. ter Set of atoms specified by global ids GIDs whose coordinates should be used to center trajectory frames Requires periodicfix to be true List glue A list whose elements are lists of GIDs each GID list is to be kept together during centering and periodicfix application Requires periodicfix to be true List of lists Whether to write a frame at the last step Boolean write_last_step 4 6 13 status The status plugin periodically outputs to the log the simulation speed in terms of simulation time per unit of wall clock time and at the end of the simulation writes a cumulative speed It s configured as shown app plugin list key key type status first ty interval ti verbose by Table 4 21 Configuration for status name description first First time for this action Time interval Time between actions Time verbose Print out running cumulative speed information Optional by default false Boolean 4 7 Configuring optional sections Certain sections of the configuration can be omitted entirely One such section is the profile section which can sometimes be useful for debugging and tuning performance 4 7 1 profile Each Desmond application can generate a runtime profile of time spent in various activities output at the end of the run thus helping you to tune your simulation for best performance These activities usually correspond to functions
237. the simulation according to the Boltzmann distribution for a specified temperature something you may wish to do once at the start of the simulation On the other hand trajectory writes all particle positions to the specified output file at specified intervals which you probably wish to do more than once but less often than at every timestep The main loop plugins are configured in the section of the configuration named after the application being run e g mdsim or remd Not all plugins are active in the main loop Some plugins provide integrators and additional force terms They are either partly or wholly configured in these sections of the configuration Plugins provided with Desmond are described in Configuring the built in plugins Desmond already has most or all the functionality required for typical molecular dynamics simulations but you can extend this functionality by writing your own plugins to for example support new force field terms add new integra tors or apply arbitrary steering forces to the simulation all without recompiling the Desmond executable Implement the functionality you need as a plugin then specify the parameters for your plugin in the configuration file Other requirements are discussed in Extending Desmond 14 Chapter 3 Key Concepts CHAPTER FOUR RUNNING DESMOND This chapter explains the basics of working with configuration files describes how to invoke the various Desmond applications including
238. the size of the file metainfo path to metadata frame file time frameindex Time associated with the frameindex th frame times numpy array of times associated with all frames rewind time For a writeable frameset truncate any times after time nearest time Return frame object with associated time x where x time is minimal le time Return frame object with largest associated time x where x lt time 1t time Return frame object with largest associated time x where x lt time ge time Return frame object with smallest associated time x where xz gt time gt time Return frame object with smallest associated time x where x gt time push_back frame t Append frame to a writeable frameset Time must be greater than previous last time entered meta Get the metaframe On writeable framesets changes here will be flushed to disk on closing 13 4 Python tools for trajectories and framesets 115 Desmond Users Guide Release 3 4 0 0 7 Table 13 2 Frame attributes and methods attribute or method description __ labels_ list of all field names _ endianism__ endianism of the data in this frame integer _ machineEndianism__ endianism of this machine integer __ Sameendianism__ true iff endianism of this frame matches machine endianism Boolean __has__ fieldname true iff frame has a
239. the three hydrogen atoms connected to a carbon atom in a methyl group and e three particles arranged in a rigid triangle such as a water molecule These constraints are described in detail in Constraints 3 7 Dynamics 11 Desmond Users Guide Release 3 4 0 0 7 When you prepare your structure file you specify the types of constraints if any and the atoms involved in them When you configure your simulation you can specify how precisely to compute the constraints Whether and how to use constraints depends on simulation specific factors or the force field you re using 3 8 Using Desmond Desmond is a suite of computer programs It uses a standard format for input structure DMS files and an open format for output trajectory files or frame files So you can also use other applications with Desmond both public domain and commercial 3 8 1 Input Desmond requires two files for input a structure file that defines the chemical system and a configuration file that sets simulation parameters The structure file specifies what to simulate the initial state of the system the size of the global cell the particles it contains their positions and other properties the force fields to employ and possibly other details Structure files are also called DMS files file name suffix dms for DESRES Molecular System The configuration file specifies how you want to simulate the chemical system the reference temperature an
240. thout regard to conflicts of law principles JICENSEE acknowledges that x DESRES may enter into agreements with one or more third parties each an Independent Commercial Distributor to distribute the SOFTWARE for commercial use y as of the date of this Agreement DESRES has entered into one such agreement designating Schrodinger LLC as an Independent Commercial Distributor and z any such Independent Commercial Distributor including without limitation Schrodinger LLC is a third party beneficiary of this Agreement The exclusive venue for any action relating to this Agreement shall be the state and federal courts situated in the State of New York County of New York and each party expressly consents to the jurisdiction of such courts This Agreement constitutes th ntire agreement between the parties and supersedes all prior agreements written or oral relating to the subject matter hereof This Agreement may not be modified or altered except by written instrument duly executed by both parties If any provision of this Agreement is deemed to be unenforceable that provision shall be enforced to the maximum extent permitted to effect the parties intentions hereunder and the remainder of this Agreement shall continue in full force and effect The failure of either party to exercise any right provided for herein shall not be deemed a waiver of any right hereunder 20 2 Licensed Companion
241. tion reference Desmond Users Guide Release 3 4 0 0 7 Return returns its argument a This operation stores its second argument under the name given by the first argument The storename must be declared in a static statement of the enhanced sampling configuration and the length of the array a must be the same as the length declared in the static statement Note that the store does not actually occur until the end of the potential evaluation so that the stored value is not accessible until the next potential evaluation sum Class Normal Arguments a array Return the sum of the elements of a time Class Normal Arguments None Return the chemical time for this step 153 Desmond Users Guide Release 3 4 0 0 7 154 Chapter 19 Enhanced sampling function reference CHAPTER TWENTY LICENSES AND THIRD PARTY SOFTWARE 20 1 Licensing Desmond for Non Commercial Research Desmond can be licensed at no charge for non commercial use subject to the following license conditions The terms of the license below are as of the time this document was prepared but is subject to change Consult the terms of the license agreement you obtained with your distribution DESMOND LICENSE AGREEMENT FOR NON COMMERCIAL RESEARCH 1 License Grant Subject to the terms and conditions of this license agreement the Agreement D E Shaw Research LLC DESRES grants to LICENSEE a limited
242. to give good performance in most cases The default setting of transform auto uses a heuristic method to set the value according to the above advice but the user is still responsible for ensuring that this selection is optimal for his situation Almost never should it be required to set keep_nyquist true since the amplitude of the farfield electrostatics should be small at the Nyquist frequency and if it is not signals a problem with the configuration of the simulation Additional parameters particular to the method type are also specified in this configuration section as described below Table 7 19 Configuration for far name description type Type of Ewald summation method to use gse pme n_k Number of fourier mesh points along each global cell axis List of Integers gt 0 transform c2c complex to complex transform r2c r2c 2round variants of real to complex transform Optional by default auto c2c l r2clr2c_2round auto keep_nyquist If true keep Nyquist value in transform default is false Boolean Both PME and GSE nonbonded far computations are in the far_terms Hamiltonian category 7 4 1 Particle mesh Ewald Particle mesh Ewald computations are configured as shown in force nonbonded far type pme common options order Og Oy Oz For PME point charges are spread to the mesh by convolving them with cardinal B spline functions scaled to the mesh dimensions in real space
243. ton _NPH dynamical system is R p m A n W qi si W Pi Vr U r B s 1 1 No A n p W DONE 1 a P r p B s Po B s B s 7 Y Biwi m I i where Po B Po Tr TB 1 TB with Po given by the pressure P_ref parameter and 0 0 0 T 0 0 0 0 0 t33 9 6 Dynamical systems 79 Desmond Users Guide Release 3 4 0 0 7 with t33 given by the pressure tension_ref parameters The barostat mass W is given by W 3N d k BTA where N is the number of constraint terms or molecular groups of the system and d is the number of independent 7 variables This system is not thermostated so the roles of 7T and 7 are redundant for this system However other systems use the same barostat framework and do apply a thermostat to the barostat This system preserves the scalar H r s p 7 Ho r B s p 9 oni 2W Po Tr TB s B s and the phase space density by isotropy type isotropic Qp s ds idMm No semi isotropic d Qp 1 II dsjdnjQo j l anisotropic a Qp dsjdnjQ j 1 constant area Qp ds dm Qo Like a V_NVE simulation the exact trajectory if ergodic is expected to sample from a surface of constant H r s p n weighted by Qp Table 9 8 Configuration for Piston _ NPH name description tau Used to set the mass barostat Time gt 0 barostat T_ref Equilibrium temperature used to set the mass Optional defaults to the global r
244. uations of motion can be used to control certain macroscopic parameters of the system for example the volume of cell or the temperature of the particles This section reviews basic mechanical and statistical concepts of particle motion later sections describe these different kinds of dynamics 9 1 1 Particles The basic data describing each particle are 1ts position and momentum vectors r and p and a set of usually fixed particle properties ranging from the parameters of certain particle interactions charge mass van der Waals radius to discrete parameters indicating membership in some group or another for example this particle is part of a ligand and this particle is not Given a set of particles the kinetic energy is N K p 2 I2 l 2m gt where m is the mass of the particle A force field refers to a potential energy function U r U F y which makes the total energy E of the particles N E r p l p 11 2m UG i 1 A basic problem of molecular dynamics is the time integration of the Newton equations of motion Ti P mi Pi Vr U r Fi r whose exact solutions conserve E r p In Desmond particles are placed in the global cell with periodic boundary conditions This means that long range interactions for example electrostatic interactions are in principle summed over all periodic images of the global 69 Desmond Users Guide Release 3 4 0 0 7 cell making the potential
245. ucture file for Free Energy Simulations 137 Desmond Users Guide Release 3 4 0 0 7 Table 18 4 CT level MMFEPIO properties property name description s_fepio_name Arbitrary name used to distinguish the original perturbed pair from other pairs i_fepio_stage 1 for the original ct 2 for the perturbed ct The fepio_fep block The perturbed CT has an fepio_fep block to indicate how atoms and interactions map from the original_ct onto the perturbed_ct The top level properties in the fepio_fep block are shown in fepio_fep properties Table 18 5 fepio_fep properties property name description s_fepio_name Should be the same as the s_fepio_name used in the original_ct i_fepio_stage Normally 1 indicating transformation from the ct with s_fepio_stage 1 to the ct with s_fepio_stage 2 Inside fepio_fep block are the following blocks e fepio_atommaps e fepio_bondmaps e fepio_anglemaps e fepio_dihedralmaps e fepio_exclmaps e fepio_pairmaps fepio_atommaps This indexed block maps the alchemically transformed atoms Specifically it maps the atom number from the origi nal_ct onto the perturbed_ct Table 18 6 fepio_atommaps properties property name description i_fepio_ai The atom index in the original_ct being mapped i_fepio_aj The atom index in the perturbed_ct being mapped For atoms in the original_ct i_fepio_ai that map onto dummy atoms in the pertu
246. up The potential energies are broken down by column according to their interacting pairs of groups and by row their Hamiltonian category Table 4 12 Configuration for energy_groups name description first First time for this action Time interval Time between actions Time name The output file name String options Whether to print pressure tensor pressure_tensor the correlation energy corr_energy and or the self energy self_energy List of strings 30 Chapter 4 Running Desmond Desmond Users Guide Release 3 4 0 0 7 4 6 5 compute_forces The compute_forces plugin Writes a per particle listing of forces to an output trajectory frameset p It is primarily useful for diagnostics app plugin list key key 4 type compute_forces first tf interval ti name p mode m Table 4 13 Configuration for compute_forces name description first First time for this action Time interval Time between actions Time name The directory name of the trajectory frameset String mode append noclobber clobber same as Configuration for trajectory String 4 6 6 eneseq The eneseq plugin Writes energy temperatures pressures and other summary data to an output file Configuration information is given in app plugin list key key type eneseq first tf interval t sync_io bs name
247. which to write the energy estimates Filename 4 6 3 e_bias The e_bias plugin applies a constant electric field with the direction and magnitude given by E_applied The schedule subsection can be set to none if so the field remains constant over time otherwise it s scaled by the values given in schedule value force term list key key type e_bias E_applied Ez Ey Ez Applied field in kcal mol A e schedule time ti t2 tn Times in picosecond value Si So Sy other force terms With e_bias a particle carrying charge q experiences a force gt F qEs t qEy t qE t where the electric field at time t is given by E t Ea S t fora x y z The time dependent scaling factor S t is determined by the schedule If schedule none then S t 1 for all t 4 6 Configuring the built in plugins 29 Desmond Users Guide Release 3 4 0 0 7 Otherwise S t at time t is given by piecewise linear interpolation Sit Sigr Si Ift ti tia S 4 S ift lt t 4 3 SN ift gt ty e_bias is often used to model electric potentials across membranes Table 4 11 Configuration for e_bias name description E_applied Applied electric field in kcal mol A e List of 3 Energy Length ElectronCharge schedule time Times at which scale factors are specified List of Times schedule value Scale factors to apply at corresponding time
248. wing we will frequently refer to processor cores as processors where confusion is unlikely When you run Desmond in parallel specify the number of Desmond processes you want to run according to the particulars of your parallel environment You can run Desmond in parallel that is run multiple Desmond processes and also run each process with multiple threads using the t pp command line parameter In order to run Desmond in multi threaded mode efficiently you ll need to request as many total processor cores as the total number of threads For example if you are running on a system with 8 processors cores per node and specify 2 processes per node then you should set the t pp parameter no larger than 4 The details of selecting the number of nodes and processes per node are system dependent and are not discussed further When running a simulation in parallel Desmond processes exchange the information by means of a parallel communication interface typically MPI implemented with a plugin called a destrier That implementation is registered under a symbol normally either mpi or serial by which it can be selected by giving an application the destrier flag desmond destrier mpi tpp 1 includ xample cfg Without the destrier flag a Desmond application defaults to serial The details of Desmond installations and parallel environments vary but a plugin containing a destrier implementation in a file named destrier so and registere
249. y and proprietary rights inherent therein or appurtenant thereto in all media now known or hereinafter developed and LICENSEE shall protect the foregoing to at least the sam xtent that it protects its own confidential information but using no less than a reasonable standard of care LICENSEE is not purchasing title to the SOFTWARE or copies thereof but rather is being granted only a limited license to use the SOFTWARE only in accordance with this Agreement LICENSEE shall not use DESRES or its affiliates or licensors names or marks or employee names or adaptations thereof in any advertising promotional sales or other materials without the prior written consent of DESRES or if and as applicable of DESRES s affiliates or licensors LICENSEE shall inform DESRES promptly in writing of any actual or alleged infringement of DESRES or its licensors rights and of any available evidence thereof 7 Term and Termination LICENSEE s license with respect to the SOFTWAR shall be perpetual subject to DESRES s rights to terminate this Agreement Any and all rights granted to LICENSEE hereunder shall terminate immediately upon LICENSEE s breach of or non compliance with any provisions of this Agreement In the event of any termination of this Agreement for any reason LICE
250. y remd and triggers plugins to act on those frames Configuration information is shown in vrun title w plugin sas input bootfile frameset frameset name p first tf interval ti last_time t Loads a configuration or sequence of configurations given by the set of frames from a trajectory file p is expected to be a path to a frameset a trajectory output If not given then the initial configuration is processed as loaded Table 4 7 Configuration for vrun name description plugin See Using plugins configuration title A string to be included in various output files Optional by default no title string input Input mode either frameset or bootfile String frameset name Path to the input trajectory Optional filename frameset first Start processing frames after this chemical time time frameset interval Skip this much chemical time between frames time frameset last_time Stop processing after this chemical time time 24 Chapter 4 Running Desmond Desmond Users Guide Release 3 4 0 0 7 4 5 Naming output files Output files are created according to a format string having terms that are expanded on a per file basis These terms are of the form X where X is a single character they expand as listed in Terms for naming output files Table 4 8 Terms for naming output files Term Expands to B A b
251. y system ndir1 is the number of directories at the top level while ndir2 is the number of directories at the second level For typical framesets these numbers are 0 and 0 i e framefiles are stored directly under the top level directory Desmond frames contain the following fields 111 Desmond Users Guide Release 3 4 0 0 7 FORMAT char WRAPPED_V_2 FLT float or DBL_WRAPPED_V_2 FLT double CREATOR char DESMOND VERSION char Desmond version ELAPSED double wallclock from start TITLE char Title from configuration PROVENANCE char Build source info BUILDCLASS char real or double will match FORMAT KERNEL char e g linux PROCESSOR char e g x86_64 ISROGUE uint32 1 for releases O for internal builds CHEMICALTIME double simulation time in picoseconds ENERGY double in kcal mole POT_ENERGY double in kcal mole KIN_ENERGY double in kcal mole EX_ENERGY double in kcal mole FORCE_ENERGY double in kcal mole TEMPERATURE double in Kelvin VOLUME double in cubic Angstr ms PRESSURE double in Bar PRESSURETENSOR double 9 in Bar TEMPERATURE_PER_GROUP double ngroups in Kelvin DEGREES_OF_FREEDOM double dimensionless DEGREES_OF_FREEDOM_PER_GROUP double ngroups dimensionless CHARGE_SUM double electron charge CHARGE_SQUARED_SUM double electron charge squared POSITION FLT 3 natoms in Angstr ms V

Download Pdf Manuals

Related Search

Related Contents

Addtron Technology KE1020A Telephone User Manual  HP LE2002xm User's Manual  Istruzione per l'uso  Samsung FH18ETX1 User Manual  PS600 BADU Top 12 MANUAL DE INSTRUCCIONES DE  DVI-D KVM Manual  Web Server VIMAR By-web Installationshandbuch  添付品 仕 様  HP OfficeConnect Router Series Installation Manual  Manuel d`installation (800/1100/1500)  

Copyright © All rights reserved. Failed to retrieve file