COSMOsim3D and COSMOsar3D
Contents
1. ccf fcos file as used in the input file or on the command line WRTLSMA writes all local sigma moments to one file named lt inputfilename gt 1lsp where inputfilename is the name of the input file NOLSANAM switches off writing of compound names in 1sp and 1sm files Applies only with one of WRTLSP WRTLSPA WRTLSM Or WRTLSMA WRTSMS triggers writing of a local sms file The sms file is required for visualization of molecular similarity with COSMOview COMPRESS ZIP GZ triggers writing of 1sp 1sm and sms files in ZIP or GZ format OUTFIL triggers writing of the translated rotated geometry as CS3D xyz OUTFIL xxx sets the filename for the results By default the format is xyz The extension pdb is optional Self consistent alignment mode NSC number of target compounds used in the self consistency cycle 1 NSC gt 1 triggers the self consistent alignment mode NSCSTART number of compounds used to start self consistency cycle 1 NSCSTART gt 1 triggers the super self consistent alignment mode DELSCCONV convergence criterion for average similarity in self consistency mode 0 01 KSCMAX Maximum number of self consistency loops 5 COSMOsar3D Molecular fields for COSMOsar3D are written with keywords WRTLSP WRTLSPA and WRTLSM WRTLSMA and can be used for subsequent PLS analysis The NOLSANAM keyword switches of writing of compound names in 1sp and 1sm files so that the files can be used directly for PL2. ignored Example input file for a super self consistent alignment using 3 target molecules where each of the target molecules is used as start molecule for a self consistent alignment cycle A local sigma profile file without compound names is written only with COSMOsar3D license NSC 3 NSCSTART 3 NRAND 100 WRTLSP NOLSANAM fdir compounds mol _18 cosmo THIOL 14 cosmo mol 09 cosmo COO_23A cosmo SQ29852 2A cosmo SQ29852 2T cosmo THIOL 20A cosmo With input file file inp COSMOsim3D is started by Linux systems cs3d file inp Windows systems cs3d exe file inp Pair wise similarity The similarity of pairs of cosmo files can be determined on the command line The general syntax is Linux systems cs3d filel file2 KEYWORD Windows systems cs3d exe filel file2 KEYWORD filel target compound cosmo or fcos file file2 probe compound cosmo or fcos file KEYWORD is optional If no keywords are given the default values will be used For a list of keywords refer to section Keywords Example for the execution of COSMOsim3D with cosmo files for two compounds on the command line gt cs3d benzene cosmo toluene cosmo STATUS cmpd sc cycle ssc cycle 1 1 1 STATUS cmpd sc cycle ssc cycle 2 1 1 final similarities in grid 0 737 0 817 0 707 0 833 0 942 0 833 9 9 benzene toluen inp toluene cosmo COSMOsim3D stops after Keywords 2 compounds List of keywords wit
3. COSMOsim3D and COSMOsar3D User s Manual COSMOlogic GmbH amp Co KG Imbacher Weg 46 51379 Leverkusen Germany Intro du O A ieee wee Gee 3 PESQUISE o oi 4 Technrcald tarls A A ad 4 ON 7 R nnine COSMOSMI D A ias 7 Wi Opus A A Ads 7 PA N LA EA i 8 A T E E E E E E E a eaehaes 9 General E nth cdanes Uiguantesbedsa tial dese ndhedeeds peated aeasdte 9 Programi COMO ii EI sees Dassen load aes 9 Random loop CORTO Li AS 9 A E R 10 Self consistent alienment Mode medina doi 11 COSMO sar D usina aiii 11 a tated tate esta eae 11 Pra ds 13 COSMOview visualization ooooncncncncnnonnnnnnoncncnnnnononananacocnnonononnnninorononnnnonnnnanororornononananininess 14 Introduction The Conductor like Screening Method for Realist Solvation COSMO RS has become an efficient and versatile tool for the prediction of a broad range of fluid phase thermodynamic properties based on quantum chemical calculations for solutes and solvents In the framework of COSMO RS the COSMO surface polarization charge density o and its molecular surface histograms the o profiles have been proven to be excellent descriptors for the quantification of the most important kinds of molecular interactions in the liquid phase such as polar interactions hydrogen bonding and hydrophobicity The superior suitability of o for the quantification of hydrogen bond interactions has been further confirmed in a recent quantum chemical study Since the same intermolecular i
4. ORDERCE WRTL SP triggers printing of random loop results triggers writing of random loop probe geometry triggers writing of a translated rotated probe COSMO file applies only for wrtcos keyword triggers writing of one SDF file for all compounds new coordinates are replaced in the template xxx By default COSMO files and molecules in the SDF or MOL2 template are matched by coordinates Structures in the template sd file should be the same as those you want to write to the new sdf Also the order of the atoms in the sd mol2 file should be identical to the order of atoms in the cosmo file Hydrogen atoms are required to be present in the template sd file If no match is found for the atoms then COSMOsim3D tries to match by molecule names triggers writing of one MOL2 file for all compounds otherwise identical to WRTSDF triggers writing of molecules in the SDF or MOL2 file in the same order as in the xxx template rather than in the order of input COSMO files Requires WRTSDF Or WRTMOL2 writes a local sigma profile file named lt compoundname gt 1sp where compoundname is the name of the cosmo ccf fcos file as used in the input file or on the command line 11 WRTLSPA writes all local sigma profiles to one file named lt input filename gt 1sp where input filename is the name of the input file WRTLSM writes a local sigma moment file named lt compoundname gt 1sm where compoundname is the name of the cosmo
5. OSMO files with the extension fcos generated by COSMOquick or COSMOfrag can be used CF COSMO files provide approximate COSMO o surfaces based on the 3D geometry of the molecule and the COSMOfrag database of precalculated o profiles For ionic compounds CF COSMO files for different conformers can be generated from a COSMO file of one conformer of the molecule COSMO and CF COSMO files contain information about the positions Xi yi Zi the areas ai and the COSMO polarization charges qi for all COSMO surface segments Technical details In one dimensional COSMO RS methods like COSMOtherm or COSMOsim a locally averaged polarization charge density oj is calculated for each segment i according to the standard procedure using an averaging radius of 0 5 A Standard o profiles are generated as histograms with o bin width of 6 0 1 e nm from all segments of a molecule For the generation of a o profile the area of each segment i is associated to the two neighboring o grid centers of the actual value oi denoted as oj and oj according to o distance weights Wi 0 0 8 1 w 0 o _ 6 This weighting generates a smooth and charge conserving assignment of the COSMO surface segments to histograms ensuring that the integral of the o profile is the total surface of the molecule and that the o weighted integral i e the first moment of the histogram is the sum of the original COSMO charges 1 e the negative of the total charg
6. S in the open3dgsar program Output In general all output is written to the working directory except when writing to a subdirectory is triggered with the TDIR keyword If COSMOsim3D is run with input file file inp an output file file out will be written in addition to the standard output which is printed to the screen The output file starts with information about the program version and license The keywords from the input file are printed to a separate line In the case of a super self consistent alignment the name of the start 12 compound yielding the best averaged similarity of the target compounds is also printed to a separate line In the following one line per molecule is printed to the output file e number of the current compound in the compound list of the input file e six different similarity values per compound SMS3D 3D sigma similarity as calculated by COSMOsim3D eq 1 ARGETSMS3D target focussed 3D similarity MS1D global molecular similarity O 10 207 0 0 T PROBESMS3D probe focussed 3D similarity S i AR Q ETSMS1D global target focussed molecular similarity O PROBESMS1D global probe focussed molecular similarity e integer value indicating the number of the random start orientation leading to the optimal alignment e integer value indicating the lowest number of a random start orientation leading to an alignment with an SMS value within a certain threshold of the best sms3D value found
7. cy and intensive colors while areas with lower similarity appear more transparent so that the underlying molecular structure is visible 15 b ele Similarities 1 000 1 000 1 000 5 1 0 615 0 702 0 577 The following options can be selected from the menu bar OPEN Open one or more files from the file system File types can be selected from the pull down menu To visualize a sms file select FILES OF TYPE cosmo file wr1 cosmo fcos ccf sms If several files are selected and opened COSMOview will automatically generate an appropriate multiframe window SAVE amp Graphics can be saved optionally with transparent background RESET CAMERA Reset the camera to its initial position LINK CAMERA 4 In a multiframe window apply camera movements rotation and zoom to all frames By default cameras in a multiframe window are linked ARRANGE ALL WINDOWS F The arrangement of multiple COSMO view windows can be changed SETTINGS Change color labels atom settings bond settings and so on DISPLAY SIGMA SURFACE If a surface is loaded it can be hidden and shown again 16 DISPLAY MOLECULE 4 The molecule structure can be hidden and shown again DISPLAY WIREFRAME Instead of closed object surfaces it is possible to show only the wire frame This o
8. dentical to the sum of the total surface areas of target and probe The SMS sigma match similarity calculations are performed according to reference 2 with the default parameters derived in that paper A target focussed or probe focussed similarity can be calculated by using the area of the target o profile or probe o profile respectively for normalization Starting from a number of determined and random orientations the position of the probe molecule is optimized until maximum similarity is reached This is done by a trial and error line search in the direction of each of the 3 unit translation and unit rotations with minimum steps of 0 01A and 0 1 respectively After each translational or rotational step the SMS3D is re evaluated and the step is accepted if it leads to an increase of similarity The first 21 starting points in this procedure are determined the following are random orientations In order to achieve maximum performance of COSMOsim3D routine for each start position of the probe the optimization of the probe vs the target is first performed on a grid with 3 fold grid distance which reduces the number of grid points for the SMS evaluation by a factor 27 After rough convergence on this crude grid the optimization is continued on the fine grid if the achieved similarity is higher than a certain percentage 70 of the maximum similarity achieved on the crude grid before and if the optimized position is sufficiently distant fro
9. e of the molecule Instead of generating just one such one dimensional o profile for the entire molecule in COSMOsim3D a local one dimensional o profile is generated at each grid point of a regular 3D grid i e a 4 dimensional histogram with three Cartesian dimensions x y z and o as fourth dimension For the 16 neighboring grid points of a segment i with coordinates Xi yi Zi 0i in the 4D space the weights w w w w are computed in complete analogy to eq 1 and the segment area is assigned to the 16 neighboring grid points according to the product of the four weights This weighting ensures a smooth linear cross over of the segment assignment if a surface segment is moved between the grid points In the first step of the COSMOsim3D procedure the target molecule is moved to the center of the cubic grid followed by the calculation of the local sigma profiles sp3d1 on the grid points Next the probe molecule in question is centered and the local sigma profiles sp3d2 on the grid points are calculated Then the 3 dimensional o similarity SMS3D is calculated as a weighted sum over the o similarities of the local o profiles according to Y al ix iy iz a2 ix iy iz sms sp3dKix iy iz sp3d2ix iy iz SMS3D sp3d1 sp3d2 ix iy iz al ix iy iz a2 ix iy iz 2 where al ix iy iz and a2 ix iy iz are the total intensities of the local target and probe o profiles at the grid point respectively The denominator is i
10. e sc comp for target molecules in grid mol_out for probe molecules indicates if the molecule is inside or outside the 3D grid e file name of the current molecule The standard output usually printed to the screen contains a status line and a similarity line per compound Entries in the status lines are the number of the current compound and the numbers of the self consistency cycle and super selfconsistency cycle Entries in the similarity lines are e in grid mol_out indicates if the molecule is inside or outside the box e six different similarity values per compound as in the output file e integer value indicating the number of the random start orientation leading to the optimal alignment e integer value indicating the lowest number of a random start orientation leading to an alignment with an SMS value within a certain threshold of the best sms3D value found e Input file name e File name of the current probe molecule e Keywords from the input file In the case of a self consistent or super self consistent alignment the standard output written to the screen contains a protocol of the complete alignment procedure There are several blocks corresponding to the cycles in the self consistent run The self consistency procedure is finished when a line stating self consistency achieved is printed After the self consistency is achieved the compounds which are not part of the self consistency cycle i e the probe molecules are ali
11. e this deficiency the molecular o surface is projected onto a 3D grid in COSMOsim3D yielding local o profiles on the grid points Arranging molecules with maximum overlap of the local o profiles on the grid leads to optimal alignment and COSMOsim3D similarity Based on the experience with one dimensional COSMOsim COSMOsim3D can be expected to enable scaffold hopping and structure independent alignment in a natural way since the alignment is based on the local molecular o surface instead of the molecular structure The array of local sigma profiles LSPs can not only be used for alignment and as a similarity measure as in COSMOsim3D but also as a novel set of molecular interaction field MIF descriptors which is optimally suited for comparative molecular field analysis e g to generate a ligand based model for the prediction of pKi COSMOsar3D is an extension of the COSMOsim3D method which uses the local sigma profiles of a set of aligned ligands as descriptors for 3D QSAR analysis Note that in this documentation COSMOsim3D and COSMOsar3D are documented together since both methods are implemented in a single software However some options require a license extension for COSMOsar3D and are therefore restricted Prerequisites COSMOsim3D requires COSMO o surfaces of target and probe molecules COSMO files can be obtained from quantum chemical calculations using quantum chemical programs with suitable COSMO implementations Alternatively CF C
12. gned and the similarities to the averaged sigma profiles from the self consistent procedure are printed 13 Additional output files are _tr cosmo files The tr cosmo files contain the aligned translated and rotated molecules i e structures and sigma surfaces Writing of the tr cosmo files can be triggered explicitly with the wRTCOS keyword Since some other options require the tr cosmo files for further processing the _tr cosmo are also written without the wRTcos keyword if another option e g WRTLSPA requires this sms files and_tr sms files Writing of the sms files is triggered by the wrTsms keyword The sms files contain the local similarity information in a way which can be visualized with COSMOview The local values of the COSMOsim3D similarity are used to control the COSMO surface transparency By default regions with high similarity are shown with low transparency and thus have intensive colors while areas with low similarity get high transparency so that mainly the ball amp stick structure of the molecule appears Note that visualization of molecular similarity with COSMOview requires tr cosmo and either sms Or_tr sms files _all sms files In the case of a self consistent alignment using more than one target compound werTsMS will also write a file named file all sms where file inp is the name of the input file In this file the averaged similarity of the target molecules is used to control transparency of all m
13. h default values in some applicable with COSMOsar3D license only General LICENSEDIR xxx FDIR XXX Program Control TARGETFOCUS ROBEFOCUS MSEXP xxx INI D xxx IDS xxx ELSIG i R RIDLSP xxx D R OT xxx TRANS XXX LOW INV P S M G G I D D A SETORI2FIRST TDIR H DIR xxx sets path for the directory containing the license file License ctd Note that the path has to be set in or If omitted the license file is searched for in the PWD and HOME directories sets path for the directory containing the cosmo fcos files of the compound list sets target focussed SMS3D for optimization sets probe focussed SMS3D for optimization raises the SMS values calculated on the grid points to the power of xxx sets minimum 1D similarity to continue 0 10 sets grid spacing in Angstrom 1 00 sets LSP LSM grid spacing in Angstrom 1 00 scaling factor for sigma intervals in LSP files 6 sets angle resolution in degree 0 10 sets translational resolution 0 01 enables inversion steps may destroy stereo chemistry moves the origin to the center of the first compound Applies only when fields are written with WRTLSM WRTLSMA WRTLSP WRTLSPA keywords triggers the creation of a cs3d_YYYYMMDD HHMMSS Xxxxxx folder in the current working directory all output files will be stored there allows for selecting a specific xxx folder for output files which will be c
14. llation directory To run this alignment job call cs3d on a shell with gt the screen output is redirected to a file e linux systems cs3d thr align inp gt thr align log e windows systems cs3d exe thr align inp gt thr align log After cs3d has finished the following files in addition to the other example files should be present alongside thr align inp thr align out the output file thr align log the redirected screen output _tr sms files for all 88 compounds i e 1 tr sms 2 tr sms 3 tr sms _tr cosmo files for all 88 compounds i e 1 tr cosmo 2 tr cosmo 3 tr cosmo The sms files can be visualized with COSMOview as described below COSMO view visualization The sms files written by COSMOsim3D can be visualized with COSMOview a free molecular structure viewer COSMOview can display one molecule per window or multiple frames with molecules in a multiframe mode The multiframe mode is useful for visualizing aligned molecules in parallel Rotation and zoom can also be done for all frames in parallel The number of frames can be set with a right mouse button click on the box in the bottom right corner of the COSMOview window If sms files are loaded the window also features sliders to adjust the surface transparency based on the local sms values The local sms values are used to control the COSMO surface transparency With default settings surface parts with high similarity are displayed with low transparen
15. m previous final points of crude grid optimizations Beyond the basic pair wise alignment and similarity calculations COSMOsim3D can also be used to build a self consistent model from several diverse molecules which can then be used as a model target for the alignment of the probe molecules For such a self consistent model the first n molecules of a list of m molecules are selected as target compounds First a standard COSMOsim3D alignment for the first two molecules is carried out Then the third molecule is aligned in a way to achieve maximum similarity with the averaged local sigma profiles of the previous compounds This procedure is continued to the n th molecule After this first cycle is finished the program starts again with the first compound and re aligns it in order to achieve maximum similarity with the following n 1 molecules The same is done for the other n 1 molecules The self consistency procedure is stopped when the total average similarity in a loop has converged up to a predefined convergence criterion Finally the remaining m n compounds probes are aligned in order to achieve maximum average similarity with respect to the averaged local sigma profiles of the first n molecules the model target This procedure results in a consistent alignment of the entire set of compounds Since the self consistent alignment can be biased by the alignment of the first two molecules a super self consistent mode can be selected in which
16. nteraction modes which govern fluid phase thermodynamics also determine ligand receptor interactions it is most plausible that a o profile based description of ligand ligand similarity or ligand receptor interactions should be very promising Based on these considerations the COSMOsim method was developed which measures ligand ligand similarities using the molecule specific global o profile disregarding the spatial distribution of the polarization charge density This approach was shown to provide useful discrimination of bioisosteric and random ligand pairs especially for smaller molecules Besides speed one of the major advantages of COSMOsim is that it naturally supports scaffold hopping by using the molecular COSMO RS sigma surface instead of the molecular structure Furthermore analogy based QSPR based on COSMOsim delivers powerful models for properties that are mainly governed by isotropic interactions like logS logP logBB etc However the selective binding of ligands to receptors is known to be based on multiple strong interactions and the 3D arrangement of the interaction sites of a ligand thus plays a crucial role As a result in target dependent problems one dimensional COSMOsim tends to retrieve false positives along with true positives This results from the fact that the global o profiles do not contain any information about the spatial distribution of the polarization charge densities on the molecular surface In order to overcom
17. olecules This may be useful to identify relevant regions in molecules for binding 1sp files and tr 1sp files 1sp files contain the local sigma profiles computed by COSMOsim3D Writing of 1sp _tr lsp files per compound compoundname 1sp is triggered by the wRTLSP keyword while 1sp files for the complete compound list named file 1sp using the name of the input file file inp are written with the wRTLSPA keyword Note that when WRTLSP Or WRTLSPA keywords are used tr cosmo files will be written additionally 1sm files Similar to the 1sp files COSMOsim3D can write files containing the local sigma moments 1sm files can be written per compound compoundname 1sm with the wRTLSM keyword or alternatively for the complete compound list file 1sm with the wRTLSMA keyword Both keywords also trigger writing of tr cosmo files Example In the following an example procedure for the alignment of a data set of THR ligands is described in detail In the subdirectory examples of the COSMOsim3D installation directory a prepared input file named thr align inp can be found 14 Nsc 1 NRAND 50 WRTSMS fdir thr ligands 45 cosmo 1 cosmo COSMO COSMO COSMO COSMO COSMO With this input file an alignment using one target compound and 50 start orientations of the probe compounds is started by COSMOsim3D The compound list consists of 88 compounds which are located in the subdirectory examples thr ligands of the insta
18. ption works for surfaces atoms and bonds USE CHARGE DENSITY PICKER 4 To get an idea of the quantitative surface charge density at a given point you can activate the charge picking mode and move the cursor over the o surface A slider at the right hand side will display the charge density at the spot you are pointing on However these values can only be approximated and are not guaranteed to be entirely precise This is mainly an effect of interpolation between the reduced grid size compared to cosmo files Please also note that since COSMOview uses an internal color correction the legend produced will not be applicable to images obtained by other means than COSMOview e g third party browser plug ins DISPLAY INFO Shows a few data on grid size and charge for Movement Molecules can be moved using the mouse buttons Rotate the molecule by dragging the mouse with the left button pressed If you move the mouse quickly you can give the molecule a spin to have it turn by itself Zoom in and out with the right mouse button pressed or simply by turning the mouse wheel l A Klamt J Reinisch F Eckert A Hellweg M Diedenhofen Polarization charge densities provide a predictive quantification of hydrogen bond energies PCCP 2011 4 955 963 2 M Thormann A Klamt M Hornig M Almstetter COSMOsim Bioisosteric similarity based on COSMO RS o profiles J Chem Inf Model 2006 64 1040 1053 3 A Klamt COSMO RS From Quantum Chemistr
19. reated if it does not exist Random loop control NSC MINPS xxx number of target compounds used in the self consistency cycle 1 NSC gt 1 triggers the self consistent alignment mode sets minimum relative similarity of current optimum to continue with fine grid search for random loop 0 50 S MINGD xxx NRAND nnn NRAND 0 NRAND 1 NRAND 2 NRAND 3 NRANDO nnn 10 sets minimum geometric distance relative to previously considered optima to continue random loop 0 30 integer specifying number of random trials 9 starts from initial geometry only no random loop leaves the probe molecules unchanged ETTR dx dy dz dphix dphiy dphiz with NRAND 1 moves probe by dx dy dz vectors and rotates probe along the axis given by dphix dphiy dphiz The norm of the vector dphix dphiy dphiz determines the rotation angle in degree ETBOX sx Sy sz nx ny nz With NRAND 1 sets the box start coordinates and the number of grid points for each direction ETORI orix oriy oriz With NRAND 1 keeps the molecules and sets the origin centers first target compound and moves all others by the same shift moves the origin to the center of the first compound and leaves the compounds unchanged integer specifying start of randomized trials neglects the initial geometry 0 Print Write Options FU RINTRAN ERAN RTCOS ER S W CENT WRTS DE XXX WRTMOL2 xxx RE
20. the COSMOsim3D executable cs3d exe is located If the path is set correctly you should be able to call COSMOsim3D on the DOS shell from any directory by typing cs3d exe If the path for the COSMOsim3D executable is not set you can still run a COSMOsim3D calculation by typing in the absolute path for the executable e g C Program Files COSMOlogic COSMOsim3D binWin32 cs3d exe Running COSMOsim3D Running cs3d without options will print a help message with a list of keywords With input file Execution of COSMOsim3D with a text input file allows for alignment of a list of molecules with the first molecule in the list and also for a self consistent alignment using up to 9 target molecules A formatted input file consisting of two parts is required e Three global command lines e List of cosmo fcos files The number of target molecules is expected as the first entry in the first line with keyword Nsc and is mandatory With Nsc 1 a single molecule is used as template for the alignment For a self consistent alignment COSMOsim3D can use up to 9 target cosmo files Nsc 9 to generate an averaged target sigma profile A super self consistent alignment is done when the number of molecules used to start a self consistent alignment is indicated with keyword NSCSTART The path for the cosmo files is indicated with keyword fdir Additionally the command lines may contain any of the keywords listed below Lines starting with are
21. the algorithm uses each of the first n molecules once as a start compound for the self consistent procedure The optimal alignment 1 e the one with the maximum average alignment on the n molecules is then chosen as a model to which the remainder of the list 1 e the probe molecules are aligned Installation The COSMOsim3D and COSMOsar3D distribution contains the executable program and some examples in a zip archive Extracting the zip archive cosmosim3d zip will automatically create the following file directory tree installation directory COSMOsim3D binLinux32 cs3d binLinux64 cs3d Linux executables binWindows32 cs3d exe Window binWindows64 cs3d exe executables docs Documentation examples Example input and output files for various options To install COSMOsim3D and COSMOsar3D on a Linux or Windows system copy the appropriate executable to a local directory and set the path If using Bourne Again Shell bash on a Linux 64 bit system add to your bashrc file export PATH SPATH software COSMOsim3D binLinux64 In a Windows environment set the path from the system control From the Windows Start menue select Control Panel then System Select the Advanced tab With the Environment Variables button at the bottom of the window you get a dialogue where you can edit several variables Edit the Path variable from the lower part and add the path to the directory where
22. y to Fluid Phase Thermodynamics and Drug Design Elsevier Amsterdam 2005 4M Hornig A Klamt COSMOfrag A novel tool for high throughput ADME property prediction and similarity screening based on quantum chemistry J Chem Inf Model 2005 45 1169 1177 5 P Tosco T Balle An open source software aimed at high throughput chemometric analysis of molecular interaction fields J Mol Model 2011 17 201 208
Download Pdf Manuals
Related Search