Home

Additional file 1

1. Animate With a Normal Mode A Nn i m ati O n Plot the structure File colored by B factors Figure 5 0 The Export Module Menue Two main export features exist plain text and PDB file Text files can be a A generated containing information about the ENM that was run type parameters etc the eigenvalues or eigenvectors Plot the ENM will make a PDB file where bonds are explicitly added that correspond to the springs used in the ENM Animations take the form of a multi MODEL PDB file where each MODEL describes one frame or state If Blocks is set to a file name then a one to one mapping from ENM point to residues in the block file will be assumed The residues determined by residue ID will then move as rigid blocks based on the motion of the corresponding ENM point 5 1 Exporting Data to a Text File Select which data types you would like to export to a text file The ctrl and Shift keys can be used to select multiple options The first six eigenvalues and eigenvectors are ignored since they represent the degrees of freedom for rigid body rotation and translation Note that the eigenvectors normal modes are printed to the file in a comma separated format and each line has up to six numbers Each mode is then the concatenation of these rows Mode 1 valuel value2 value3 value4 value5 value6 value7 valuel2 Page 27 of 30 MAVEN 1 1 5 2 Exporting the ENM model to a molecular viewer This module prov
2. each separated by a coma 1 e 100 13 7 would indicate there are 100 atoms listed in the less detailed lower resolution section a cutoff radius of 13A for the lower resolution section and 7A for the more detailed section In Figure 2 47 we show an image of the mixed resolution system The PDB file used in this example is located in the examples folder of the MAVEN distribution and named 1T3R_mixedModel pdb and is also available on our web page Figure 2 47 A HIV Protease structure 1T3R with protease inhibitor TMC114 in red and protease atoms within TA of the ligand in blue The remainder of the structure is represented by C atoms and colored green B The elastic network is exported to PYMOL The default view colors the coarse grained section green and fine grained blue ENM_GUI_y1 5 s rs ation Analwaleand Figure 2 411 MAVEN Interface for Mixed Models at yeni Mo de Loading the model into MAVEN we select an ENM type astic Models Ss w n of ANM mixedcg and set the parameters to 163 13 7 to k Please submit PDB formatted files only Prepare Files tell MAVEN that we have 163 coarse grained points we a 1 want to connect them with harmonic springs if they are s memei within 13A of each other and that points within the fine ANM Select an ENM type and ae o cae corresponding parameter s grained section should be connected if they are within 7A AN inode 10337 Note that th
3. 2 173 181 PM 9218955 Lin T L and G Song 2010 Generalized spring tensor models for protein fluctuation dynamics and conformation changes BMC Struct Biol 10 Suppl 1 S3 PM 20487510 Yang L G Song and R L Jernigan 2009 Protein elastic network models and the ranges of cooperativity Proc Natl Acad Sci U S A 106 12347 12352 PM 19617554 Page 30 of 30
4. MAVEN interface and explain many of its features Figure 1 3 Figure numbers correspond to section numbers provides an annotated screenshot and points to which sections provide further explanation Video tutorials are also available at the MAVEN web site http maven sourceforge net MAVEN only accepts PDB formatted files see www rcsb org Data files of other types or formats will not be read properly Density maps can be used by converting them within the Prepare Files module see Section 2 3 3 Press any E to display Badi files aga EEEE prepare Files e E ENA w sy a 3 S p T h e S e e cted fi l e k Please submit PDB formatted files ony PrepareFiles Press to amp C DOCUME 1 Administrator MYDOCU 1 MATLAB 1 A e select a file T3R_CA pdb Select PDB File Section 3 2 eee Select an ENM type mss eeu S lt Section 3 1 E t t Nam necresthtem 50 Quick ENM inpu arameters Computation Cempiete g _Savelload p p S AN ETA ENM Models AB Save Load an m Status Points in the Network 198 Ciu z Status Wi n dow Sum Residual 2 2299 16 ENM model aj varesiauay 1167 Relative Motion 40 60 80 100 120 140 160 180 Residue Index Section 4 Section 5 Analysis Exporting Figure 1 3 Interface Overview The MAVEN interface is shown and annotated with the corresponding section number for further details about each component Note that this User s Guide was generated using the Windows build of MA
5. structure 1TRA no experimental ANISOU records were found When this is the case MAVEN will plot the ANM derived fluctuations and color them by their relative volume We see that the terminal base on the acceptor arm dominates the motion the so called tip effect which is usually lessened with the nearest atom or STeM methods 4 3 Fluctuations in internal distances ENMs allow us to efficiently compute not only mean square fluctuations but also fluctuations in internal distances This quantity describes the average fluctuation in the distance between atom i and atom j within the modes and is calculated using lt AR ARj gt 3kpT y Tt 051 213 where kg is the Boltzmann constant y the spring constant between atom i and j and T 1 7 017 where the summation of normal modes Q and square frequencies A the eigenvector and eigenvalues Page 18 of 30 MAVEN 1 1 respectively We compute the pseudo inverse because I has zero value eigenvalues and is by definition not invertible and requires the use of singular value decomposition A Plot of the relative internal distance changes lt ri j gt lt ri gt lt rj2 gt 2 eri rj gt Vi x lt 40 A o D s a 100 h 120 S W g 140 z T 50 100 150 Residue Index Figure 4 3 Visualizing changing internal distances by using the Plot Internal Distances button see Figure 4 0 We compute an alpha carbon ANM model and visu
6. 98 We choose modes 1 through 5 1 5 for the comparison When more than one mode is chosen an overlap matrix is constructed using the first selected mode 1 in this case which consists of a heatmap of pairwise dot products between the direction of each atoms motion in the given mode A second plot is made where the Mean Square Fluctuation MSF from ENM using only the selected modes here 1 through 5 is plotted with the experimental B factors from the PDB file for both groups Any status changes including errors thrown due to improper parameter syntax will be shown in the main status window see Figure 1 3 Motion Correlation Module Overlap in Mode 1 B factors Using 1 5 Figure 4 4 1 Comparing Motion Within an ENM Here we show a comparison between the motion of the N and C terminus of IT3R using only the first mode Overlap and the first 5 modes MSF calculated using 1 5 Here we have used an ANM model with a cutoff of 13A We see that the two termini are well correlated with each other in the first mode and that the MSF from ENMs is very similar to the B factors from X ray crystallography Pearson s correlation Residue Index coefficients are displayed between the eta ce tome os Group 1 ENM MSF and experimental PDB values Corr ENM1 ENM2 0 57 Group 1 PDB When the two parts of the structure Corr PDB1 PDB2 0 56 Group 2 ENM groups are of the same size the Group 2 PDB Make The Comparison C Another ENM Yth Ditterent P
7. MAVEN Motion Analysis and Visualization of Elastic Networks and Structure Ensembles Version 1 1 User s Manual Contents 1 Overview 1 1 What is MAVEN 1 2 Major Features Parsing structure files Generating an ENM Analysis of the ENM Exporting and Visualization 1 3 Interface Overview 1 4 System Requirements 2 Model Generation Prepare Files 2 1 Selecting and reading a structure file 2 2 Using the atom type selector 2 3 Uniform coarse graining 2 3 1 Alpha carbon models of proteins 2 3 2 Generate centroid positions United Atoms of proteins or nucleic acids 2 3 3 Reading electron density maps or image files and converting them to PDB files 2 3 4 Spherical coarse graining iteratively remove points that are within a cutoff distance of one another 2 4 Mixed resolution modeling see 3 2 6 for running this type of model 3 Running an ENM 3 1 Quick ENM buttons that automatically parse coarse grained models from unprocessed files 3 2 Methods and parameter inputs for each ENM type 3 2 1 GNM Gaussian Network Model 3 2 2 ANM Anisotropic Network Model 3 2 3 Distance dependent springs Connect all pairs of points and scale their interaction strength by a power of their separation 3 2 4 STeM An ENM which includes sequential bond lengths angles dihedral angles and nonbonded terms 3 2 5 Nearest Atom Make a coarse grained model where spring connections are assigned based on an all atom repre
8. VEN There may be some minor differences in the look of the interface on the other operating systems The selected file is displayed Throughout the model preparation building analysis and exporting processes the Status Window will be continually updated to keep the user apprised of what MAVEN is doing Results of computations are often displayed here as well 1 4 System Requirements MAVEN has been built independently on Windows Macintosh and Linux operating systems Be sure to download the appropriate distribution for your system keeping in mind whether you have 32 bit or 64 bit Page 4 of 30 MAVEN 1 1 architecture MAVEN is coded in a mixture of MATLAB Perl and C The MCR should be the only dependency required to run MAVEN but a C compiler may be required to install the MCR If your system does not meet the requirements for installing the MCR the installation dialog will alert you as well as provide information on supported compilers On Windows systems this will likely be Microsoft Visual Studio Express which is free in Windows Linux and Mac systems should have gcc installed as part of the operating system Our source code is also made available with instructions for compiling or using the code outside of MAVEN This provides users on unsupported OS Architecture combinations access to MAVEN in addition to providing flexibility for extending MAVEN with user made analysis routines Source code written in MATLAB will require MATLAB
9. Y OR CONSEQUENTIAL DAMAGES INCLUDING BUT NOT LIMITED TO PROCUREMENT 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 2 3 4 Spherical coarse graining For many systems we only want to consider a subset of the points available to us Typically we choose alpha carbons since they represent the location of the peptide backbone and there is thus one point per amino acid This level of coarse graining provides a straightforward correspondence between the structure and sequence Sometimes either for a simpler view of the system or for computational requirements one may wish to further reduce the number of points considered For this purpose we provide a method to spherically coarse grain a set of input points We begin with the seed point This refers to the line number of the input file if the file has been pre processed The coordinates in Structure file are looped through and all points within Cutoff of the seed are discarded The next in the order of the input file point is then added to the seed This is repeated until all points have either been discarded or are in the seed The seed is now a list of points that are all at least the Cutoff distance away
10. alize the internal distance change matrix using all modes in MAVEN The structure is visualized in PYMOL with residues colored by the alpha carbon changes A The internal distance changes upon mode deformation are computed from the above equation and visualized within MAVEN An arrow relates one of the two largest areas of change to the structure B The six residues in each chain with target changes are colored and side chains displayed C Rotated view Page 19 of 30 MAVEN 1 1 4 4 Correlated Motion between two parts of the structure Determining what the normal modes mean for the structure is not always a straightforward task and analyzing the motion of subsets of the structure can be instructive We have developed an analysis module that compares how two parts subsets of the structure groups move in one or a few modes in order to better understand what the mode motion means Note that this type of analysis is in some ways a simplification of the Overlap Matrix module described in section 4 6 but may be more meaningful for looking at the relationship between two groups as opposed to a larger number of groups This type of analysis is accessible using the button visible in Figure 4 0 In Figure 4 4 1 we show an example of comparing the two termini of 1T3R There are 198 C atoms in this structure and as such the atoms are indexed from 1 to 198 Thus the N terminal ten C s are chosen with the syntax 1 10 and the ten C terminal with 189 1
11. alysis is the ability to run a mixed resolution system What we mean by mixed resolution is that part of the system is modeled in greater detail say all heavy non hydrogen atoms while the rest of the system is modeled in less detail say only the alpha carbon atoms To run such a model construct a PDB file where the lower resolution coarser atoms are listed first and the higher resolution atoms are listed afterwards The scripts in Prepare Files can help build these two components but familiarity with a molecular viewer may be required See Examples 1T3R_mixedModel pdb in the MAVEN distribution files for an example mixed resolution file Select the completed mixed resolution PDB file as your structure file and ANM mixed as the ENM type The input parameters should then be three numbers each separated by a coma 1 e 100 13 7 would indicate that there are 100 atoms listed in the less detailed coarser section a cutoff radius of 13 Angstroms for the lower resolution section and 7 Angstroms is used for the more detailed section Parameters 1 ncoarse The number of coarse grained points in IND 2 coarsecutoff The cutoff for defining interactions within the coarse grained points 3 finecutoff The cutoff for defining interactions within the fine grained points Note that the cutoff used for connecting fine resolution to coarse resolution is the geometric mean of the two cutoff values square root of their product Page 16 o
12. an Network Model for a given set of coordinates Any points within a cutoff value are connected by uniform harmonic springs GNM returns amplitudes of motion within each mode but no directional information See Bahar et al 2 for more information PARAMETERS 1 cutoff Any two coordinates within cutoff will be connected we suggest a value between 7 and 10A for alpha carbon models 3 2 2 ANM Anisotropic Network Model An ANM model is generated from a list of points and a cutoff value Any points within a cutoff value are connected by uniform harmonic springs A basis set of motions for this system is then computed Each vector in this basis set is a Normal Mode Thus ANM returns directions and relative magnitudes of motion For more information see Atilgan et al 1 for more information PARAMETERS 1 cutoff Any two coordinates within cutoff will be connected we suggest a value between 11 and 15 for alpha carbon models 2 numeigs optional number of eigen pairs to solve for default solves for all Example Cutoff set to 13A and MAVEN will only compute the first 50 modes 7 3 2 3 Distance dependent springs powerANM Page 14 of 30 MAVEN 1 1 We consider all pairs to be in contact weighted by a power of the distance between them This function performs an all pairs ENM where spring strength is dependent upon the inverse power of the distance between nodes PARAMETERS 1 power distance dependence spring co
13. arameters AMM Thi Between Two Parts of the Structure IR 189195 correlation between groups is also displayed Page 20 of 30 MAVEN 1 1 4 5 Compare one ENM to Another Another useful feature of MAVEN is the ability to compare one ENM to another This dialog is initiated using the button visible in Figure 4 0 To begin we show a comparison between linear distance dependent springs and quadratic distance dependence in Figure 4 5 1 The overall motion is not changed but there is a shuffling of the first 6 low frequency modes between the two models This type of analysis can be used to help researchers decide what the effect of various ENM types are on their system The Cumulative Overlap is also shown This quantity captures the total overlap between 3 6 10 or 20 modes in the initial ENM model and a combination of modes from the new ENM whose parameters are designated below Note also that GNM does not contain directions of motion so this type of analysis is of less value for comparing to GNM Any status changes including errors thrown due to improper parameter syntax will be shown in the main status window see Figure 1 3 Select an ENM type and corresponding parameter s 9 In itial EN M Figure 4 5 1 Comparing one ENM to another New Parameters Here we a C210 il ae nixedcg ANM nearestAtom AHN LUNI A t Motion Correlation Module Sele are comparing the distance de
14. displayed using the modevectors script available on the pymol wiki Example 3 Alpha carbon atoms from 1YYF tetramer left and hexamer right colored by their mean square displacement computed from all ANM modes This emphasizes the need to have the full structure present when generating biologically meaningful motions Also one may interpret the highest mobility of the tetramer as the regions that are seeking other binding partners with highest motions until the full structure is formed Then the motions shift from assembly dynamics to functional dynamics Example 4 Adenylate kinase shown as a cartoon with gray bars indicating the springs from the ANM Red arrows indicate motions in the first mode drawn with the modevectors script from the pymol wiki This is an open form of the structure The dominant mode describes a motion where the open active site flap top of the current view moves down toward the body of the structure moving towards the closed conformation Page 29 of 30 MAVEN 1 1 Please also see the MAVEN publication in BMC Bioinformatics Reference List Atilgan A R S R Durell R L Jernigan M C Demirel O Keskin and Bahar 2001 Anisotropy of fluctuation dynamics of proteins with an elastic network model Biophys J 80 505 515 PM 11159421 Bahar A R Atilgan and B Erman 1997 Direct evaluation of thermal fluctuations in proteins using a single parameter harmonic potential Fold Des
15. e base protein backbone protein side chain O2 Go O3 not protein backbone cat O04 2 3 3 Reading electron density maps eee ml c2 o2 n2 ns c4 nA o4 C5 ce ne OG ny ce ng n ca co ob Figure 2 3 21 United Nucleotide Sugar and Base tRNA structure ITRA is shown in a lines representation with backbone and ribose rings colored red and the nucleotide base blue United atoms are shown as spheres A close up view is also given for one nucleotide Figure 2 3 211 United residues and side chains Using the HIV 1 Protease 1T3R united side chains are shown as red spheres and united residue positions in blue generates a united atom point for the Notice that united residues also inhibitor molecule green The function map2pdb is designed to read in an MRC formatted 3D density map and convert all grid points that are at least level in value to a PDB file which contains a pseudo atom for each density point Thus level is used as density contour All grid points with an associated density value greater than or equal to level will be made into a pseudo atom The space between each grid point will be grid in Angstroms If the density map is downloaded from a database like the EMDB www emdatabank org a suggested level will be given See Figure 2 3 37 which uses EMDB structure 1800 the HIV 1 spike http emsearch rutgers edu atlas 1800_mapparams html It is notable that the functions used will
16. e cutoff length that bridges the two is v13 7 A g Aneu G P 9 5A The result is a model with much better agreement Points in the Network 425 n with the experimental temperature factors correlation y varesiauay 283 coefficient is 0 7 The correlation between experimental and computed B factors is 0 57 for only alpha carbon coordinates and a cutoff of 13A Recalling that the first 163 atoms in the file are for the low resolution section we can see from the plot on the left that this region in particular has excellent agreement with the experimental data Relative Motion Residue Index Run Further Analysis Export to Molecular Viewer Page 12 of 30 MAVEN 1 1 3 Running an ENM One must select an input PDB file ENM type and proper parameters to run an ENM model Help files for each ENM type are available within MAVEN by pressing the button with the ENM type selected Alternatively quick ENM buttons exist that allow the user to select an unprocessed PDB file and directly run an alpha carbon model C DOCUME 1 Administrator MYDOCU 1 MATLABE i i NM GUIAT3R_CA pdb A E ka Select PDE File Select an ENM type and corresponding parameter s near Figure 3 0 Instructing MAVEN what ENM to use A Pressing the 5 lect PDB File gt button will bring up a file selection dialog Navigate to the directory where your PDB file of choice is located and select it The panel to the left
17. e number of computed modes to 100 and change the weights to 2 2 10 1 These two changes together increase the correlation to 0 74 in addition to a noticeable decrease in the magnitude of the residuals difference between the curves Page 15 of 30 MAVEN 1 1 3 2 5 ANM_nearestAtom This ANM model is much like the standard ANM but the connectivity matrix is formed from an all atom model When choosing the PDB input file select a file with for example all heavy atoms in it ANM_nearestAtom will then extract the alpha carbons and phosphate atoms and connect them with uniform springs if any member of their residue or base is within the given cutoff If the third input is given an argument that evaluates to True 1 is fine then the spring strength is set to the number of atom atom contacts that fall within the cutoff PARAMETERS 1 cutoff two coarse grained points are considered in contact if any atom in their residue or base is within cutoff distance A 2 numeigs optional number of eigen pairs to solve for default solves for all 3 weighted optional If true any positive non zero number such as 1 the spring strength equals the number of atom atom contacts between cg points If false 0 then all springs have a stiffness of 1 unweighted The default is unweighted 3 2 6 ANM with mixed coarse graining See Section 2 4 for more information on constructing these models One powerful tool in ENM an
18. es a coarse grained model but uses an atomic model for determining connectivity The Spring Tensor model expands the energy function of ENMs to account for bond and torsion angle changes Mixed coarse graining represents a method for computing modes of motion in a coarse grained system while still being able to analyze molecular effects on those motions such as residue or base mutations drug binding proline isomerization or post translational modifications A second feature of this application is the ability to handle large systems through sparse matrix methods and the ability to calculate only the lowest frequency modes Since the contribution of each mode to the total motion decreases quickly calculating only the lowest frequency modes captures the majority of dynamics while using considerably less computer resources A further benefit of this platform is that it is setup to accept protein RNA DNA and ligand coordinates One can generate a standard alpha carbon model in only three clicks from an unprocessed PDB file or our atom selector can be used to save a subset of atom types for use in any ENM points can be picked from electron density contours united atoms representing Page 3 of 30 MAVEN 1 1 the centroid of a set of atoms can be generated or one may compile an initial model using other software such as a molecular viewer and use MAVEN for ENM generation and analysis 1 3 Interface Overview In this section we will introduce the
19. eshold value 0 5 used here and positively correlated in the upper TermB triangle while the lower triangle displays the fraction of FlapB values within each overlap matrix whose magnitude is shoulderB greater than the threshold and are anti correlated To relate Par the above two overlap matrices to this summary matrix we s ror S amp amp consider the Flap A row move across to the Flap B ee es Fee Ss eS column and see a white square By the colorbar we see that about half of the values in the FlapA FlapB overlap matrix are positive and greater than 0 5 We also find information such as 1 the termini exhibit the strongest positive overlap 2 both termini have strong negative overlap with both shoulders 3 the cores are the largest subsection and have little discernable overlap pattern with any other part of the structure in this mode of motion The other type of summary matrix not shown here displays Page 25 of 30 MAVEN 1 1 only an upper triangular matrix and records the fraction of each overlap matrix that has a magnitude greater than the threshold absolute value of the overlap by combining correlations and anticorrelations By choosing appropriate subsets of points detailed information about the mode shapes can be determined 4 7 Compare modes to Principal Components of a structural ensemble The goal of most ENM studies is to determine the dominant functional motions of biomolecule
20. esponding squared frequency and c the largest mode index we wish to consider The first 6 modes correspond to rigid body translation and rotation of the whole system and so their contribution is neglected 4 2 Anisotropy of Motion This function is accessed by pressing the antea button To the right of this button are two numbers They are the scaling factors used for the two sets If the numbers are changed and the button pressed again then the ellipsoids will be redrawn with the new scaling ii ee 30 40 sp 19 9 Figure 4 2 Visualizing Anisotropy A Anisotropic temperature factors from C atoms in 1T3R are shown as red mesh ellipsoids From ANM models the concept of mean square fluctuations generalizes to a 3 by3 tensor that describes more details of the directions of distortions Anisotropy of motion is more difficult to model but may be important for function Always keep in mind what the experimental data is however If anisotropies are strongly influenced by crystal packing then low agreement with the ANM results is not necessarily unwanted A similar metric can be derived from NMR ensembles by calculating the variance of each atom across the models B We show a cartoon representation of the protein to show the orientation seen in A inhibitor colored red and active site flaps blue This orientation was chosen to emphasize the differences in anisotropy between the ANM and ANISOU records in the PDB file C In the case of this tRNA
21. f 30 MAVEN 1 1 4 Analysis Coeretation Matn from ENM If an ANM model is used and all e ey atoms have ANISOU op Lr Residue index d ar a v 3 f anes 7 SRA LS SRS ja Os records plot the ico ae Seon tN two sets as ellipsoids ee ake a plot Resdue index ANISOU Scales 21 ClSplayS motion Plot ANISOU field E correlation g Plot the change Plot Fluctuation Correlation Matrix t SO haa in pairwise distanceS __Plotinternal Distances within the structure Correlate Motion Generate Overlap Matriges ensemble and compare to modes Matrix Module Figure 4 0 Introducing the Analysis module The analysis capabilities of MAVEN and the user interface This module is opened from the main MAVEN interface see Figure 1 3 by pressing the Run Further Analysis button This figure will be referenced in the next subsections 4 1 B factors During X ray crystallography an electron density is computed for the system Thermal noise as well as internal motions of the molecule contributes to the smearing of this density Thus highly mobile atoms may contribute to a relatively large area of electron density This phenomenon is quantified in a metric known as the Debye Waller or B factor and is recorded in the PDB file Because thermal motions scale with temperature they are also referred to as temperature factors It is vital to note that all B factors can have substantial contributions of ri
22. from one another The Seed can be any MATLAB style vector I is the default but you could use 1 10 or 1 5 100 or 7 13 52 7 101 218 all of which describe valid MATLAB vectors This method helps to keep the distribution of points somewhat uniform within the volume of space that the original set occupied Coarse graining along the sequence of a protein keeping every fifth residue for example will likely result in somewhat non uniform spatial sampling possibly a defective model Since packing density and the shape of the biomolecule are strong determinates of their motions it is often preferable to coarse grain spatially rather than along the sequence 2 4 Mixed resolution modeling One powerful tool in ENM analysis is the ability to run a mixed resolution system What we mean by mixed resolution is that part of the system is modeled in high detail say all heavy non hydrogen atoms while the rest of the system is modeled in a less detail say only the alpha carbon atoms To run such a model construct a PDB file where the lower resolution coarser atoms are listed first and the higher resolution atoms are listed after see Figure 2 47 The scripts here can help you to build these two components but familiarity with a molecular viewer may be required Page 11 of 30 MAVEN 1 1 Select the completed mixed resolution PDB file as your x y z file and ANM mixed as the ENM type The input parameter should then be three numbers
23. gid body motion experimental noise errors heterogeneity in the crystal etc which can all contribute to B factors Thus it is not always clear a priori how well the B factors capture the actual solution dynamics of molecules After calculating any ENM model a theoretical B factor curve is calculated from the ENM and displayed on the main MAVEN interface along with any B factors in the input structure file see Figure 1 3 Understanding which parts of the structure are correlated in the dominant motions can be very informative The Plot B factor Correlation Matrix button see Figure 4 0 does this by computing C i j oe AR AR AR AR The C matrix then represents the orientational cross correlations between the fluctuations of ENM points Plots of C for individual modes or a collection of modes will provide information about what parts of the structure move collectively For a more in depth look at how specific regions of the structure move with respect to each other see Section 4 2 Overlap Matrices Page 17 of 30 MAVEN 1 1 The normal mode fluctuations alter the relative position of atoms within the structure This change in the relative positions of atoms can be quantified by the internal distance changes AR AR AR AR 2 AR AR l 1 These are computed directly from the pseudo inverse constructed as H71 T a i 79 Vi Qi where Q is a normal mode A the corr
24. hen points in the plot are clicked on with the crosshair up to the first 8 letters of its file name are displayed next to the point an Oo Oo 3 eke o 6 RR ao oO ta mm 10 RRA Percent Variance ian mn Number of Modes No oO 20 RE h oO oO Principal Component Figure 4 7 2 PCA analysis of 2K TD A Visualizing the structural ensemble of in PYMOL coloring is blue to red from the N terminus to C terminus There are 15 models in the ensemble B MAVEN plot showing the percent of ensemble variance that is captured by each PC as well as the cumulative C MAVEN plot showing the cumulative overlap dot product of the first 3 6 10 or 20 modes with each of the first 3 PCs In this case we have generated 200 modes from the STeM ANM model that accounts for changes in bond length bond angle and torsion angle Page 26 of 30 MAVEN 1 1 5 Exporting to Text or Molecular Viewers or Text Files Often it is useful to visualize the normal modes moving the structure or to save them for other uses Both of these options are available within MAVEN from the Export module Access this module by pressing the Export to Text or Molecular Viewer button on the MAVEN main interface see Figure 1 3 Export i fx Export to Text File oe apap Factors S e l e ct th e d ata Al Eigervahies x you want saved Export to Molecular Viewer Plot the ENMI Plot the springs and nodes Generate
25. i Conversion from Density to coarse grained pseudo atoms A The map is visualized at the 2 3 density level B After selecting EMD 1800 map as the structure file within the Prepare Files module this file is provided in the Examples folder in the MAVEN distribution we set the Grid to 3 5 and Level to 2 3 and run map2pdb The resulting PDB file is visualized as spheres It is evident that many points are generated with this method and the representation is dense C The PDB file from B is coarse grained using spherical coarse graining see Section 2 3 4 for details An approximate surface representation of the points in B is shown for comparison This more sparse representation will require a longer ENM cutoff but will be much less demanding for calculations and still provide approximately the same motions We use readMRC m which was downloaded from the MATLAB file exchange on Nov 9 2010 The function is free for academic use as long as the following legal text is retained with its distribution Page 10 of 30 MAVEN 1 1 readMRC Copyright c 2010 Fred Sigworth All rights reserved 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 MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT INDIRECT INCIDENTAL SPECIAL EXEMPLAR
26. ides options for exporting the active ENM model to a molecular viewer Which viewer PyMOL VMD etc will be used and how the representations will appear when initially loaded is defined in a configuration file See MAVEN_Export conf on our web page for examples of these configuration files The model can be made into an animation Scale 1s a scalar value describing the maximum amount of deformation along the requested mode This is usually chosen to be sufficiently large for viewing purposes The default value of 100 is a good starting point The main Status window see figure 1 3 will report maximum RMSD in the animation Step defines the step size that each frame also the MODEL number in the resulting PDB file will take If Block is a file name then this file is read and then the blocks are animated More precisely we will assume that each node in the ENM corresponds to one residue in the Block file a PDB file Each residue will then move rigidly in the same way as its point in the ENM If one performs alpha carbon coarse graining then residues will move rigidly and the animation can be displayed with cartoon representations rather than simply the alpha carbon trace Note that the Block file must have the same number of residues as there are nodes in the ENM If not MAVEN will not know which residues correspond to the modeled points Many resources exist for making interesting figures using VMD and PyMOL S
27. inA chainB Compute result Only use for this set alah 0 5 of modes apove iS MATLAB OM i shreshold Directory where J This function computes pair wise overlap all output wi Il for com pa ri ng matrices for a set of subsets of the be placed the motion of structure each part or Description Normal modes describe a vector of substructure motion We can superimpose this vector onto the structure The normal mode then describes a 3x1 motion vector for each atom If we make these 3x1 vectors unit length and consider their pair wise dot Figure 4 6 Schematic of the Overlap Matrix dialog for computing directional correlations of computed motions within the structure This function computes pairwise overlap matrices for the specified modes and parts or subsets of the structure Access this module by pressing the Generate Overlap Matrices button in the Analysis module see Figure 4 0 Normal modes describe a vector of motion which we can superimpose onto the structure The normal mode then describes a 3x1 motion vector for each atom If we make these 3x1 vectors unit length and consider their pairwise dot product we will end up with a matrix describing the amount of motion in the same direction opposite direction or orthogonal Matrices of this type are termed overlap matrices NOTE As the number of parts and requested modes increases the algorithm may generate a large number of images There wil
28. iverse PDB structures as there may be more atoms in a given file with Atom Name CA than the alpha carbons of the protein similarly for nucleotide atom types A good example is that the Qpo will likely retain phosphate atoms positions from any additional phosphate ions in the solvent or from ATP or ADP Inspection of generated PDB files will show which atoms have been used Page 13 of 30 MAVEN 1 1 3 2 Methods and parameter inputs for each ENM type A list of the ENM types available and their parameters are provided below Note that parameters are separated by and not whitespace If a parameter is a vector then spaces should separate the components of the vector i e 1 1 2 5 is a valid vector but 1 1 2 5 will be interpreted as 4 parameters not one Most ENM types allow the user to designate the number of modes to solve for We recommend not asking for fewer than 50 as some of our numerical tests have resulted in poor convergence with fewer modes The main Status window will alert the user if the eigenvalues do not converge If this happens the number of requested modes should be increased Figure 3 2 Help within MAVEN The text of section 3 2 1s available from within MAVEN by selecting the given ENM type and then pressing the button next to the parameter input box ANM STeM ANM_ mixedcg ANM_nearestAtom 163 13 gt 3 2 1 GNM Gaussian Network Model This function computes the Gaussi
29. l be one image made that compare motion within each mode and for each of the pairwise parts In addition to these summary matrices will be made one for each mode that display the dominant trends See Figure 4 6 2 for a specific example It is recommended that MAVEN not be interrupted while the computation is in progress Doing so has the potential to disrupt the image saving script Defining the regions Parts within the structure that we wish to follow takes on MATLAB cell array syntax where each cell describes one region The description of a region is a vector of node indices By node indices we mean the index of the node or geometric point as it is read into MAVEN If records other than ATOM and HETATM have been trimmed from the PDB file then there will be a one to one correspondence between node index and PDB line number Depending on how the input PDB file was generated this many not necessarily correspond to the atom numbering Page 23 of 30 MAVEN 1 1 A directory for the output must be given This module will generate a folder of data for each mode of motion containing comparisons of all unique pair wise combinations of the various Parts EXPLANATION OF VARIABLES EXAMPLES ARE FOR 1T3R pdb part e cell array where cells are vectors describing the subsets to use e each cell should be a vector of induces from the parent structure e i e 1 99 100 198 1T3R is a homodimer Our first example figure 4 6 1 si
30. ly these selected atoms The status window will display the name of the newly generated file Page 7 of 30 MAVEN 1 1 2 3 Methods for Uniform Coarse Graining The next sections describe MAVEN s methods for uniform coarse graining cg The main advantage of cg models is reduced computation time It has been shown that the motions derived from coarse grained proteins are very similar to those of atomic systems As packing density and shape are the primary properties that ENMs depend on it is important that the model points reflect that original density spatial distribution of points and shape of the structure 2 3 1 alpha carbon models Figure 2 3 1 Examples of Protein Representations A PDB file 1T3R shown as sticks This level of detail does not show any solvent atoms or any of the system dynamics yet it is complex B Structures are often shown in simpler cartoon views that emphasize the path of the peptide backbone through space C We show spheres for each C atom in the structure a common way to represent proteins in coarse grained ENMs Elastic Network Models consisting of alpha carbon C atoms are given an atom type of CA in PDB files atoms from proteins are the most common type This one point per residue level of coarse graining makes model generation fast and retains a one to one relationship with sequence information Also alpha carbons are part of the protein backbone so their position informs al
31. mply compares the dimmers names e cell array of names of the subsets e each cell is a string that names the corresponding part e i e chainA chainB 1T3R is a dimer We label each part by its PDB chain see figure4 6 1 Modes 1 6 e which normal modes to consider e the contribution of each mode to the total motion diminishes quickly as the mode index increases e between 6 and 20 modes are suggested for this type of analysis threshold 0 5 e only overlaps above this magnitude will be counted in the summary matrices for both positive and negative correlations outdir the output directory e directory where the output is deposited e subdirectories will be made for each mode and all unique overlap matrices placed therein e each mode will also be summarized in two heatmaps see Figure 4 6 2 Example 1 HIV Protease 1T3R Parts 1 99 100 198 Names chainA chainB Page 24 of 30 MAVEN 1 1 Figure 4 6 1 Example of overlap matrix The result of this computation is a set of images and data files placed in the output directory This type of matrix will be referred to as an Overlap Matrix and each point in the heatmap is the overlap dot product between the directions of motion that each point moves for the given mode mode 1 shown A value of 1 means that the two points move parallel to each other in the mode and 1 is motions in opposite directions or anti correlation of mo
32. nstants are nee where 7 is the Euclidean distance between nodes i and j We suggest a power of 2 for proteins 2 numeigs optional number of eigen pairs to solve for default solves for all Example Generating 100 modes with spring strength set to rj ias NOTE Rather than the parameter for the cutoff distance we now have a parameter for the power dependence For more details including a performance comparison with cutoff based ANM see Yang e al 4 3 2 4 STeM Spring Tensor Model This is an ANM model that accounts for three and four body interactions a Spring Tensor Model See the recent paper by Lin and Song 3 PARAMETERS 1 neigen number of eigen pairs to solve for We recommend solving for 50 or more 2 weight optional the weights used to combine the 4 hessians bonded pairs bond angles tortion angles and nonboned interactions Default is 1 1 1 1 NOTE Only use spaces for separating the weights Commas are used for separating parameters Example Running STeM with 100 modes with changes to bond length and angles weighted twice as strongly as nonbonded and torsion angle changes weighted by ten 100 2 2 10 1 B Factor 0 20 40 60 a0 100 120 140 160 180 Figure 3 2 4 Effect on STeM performance with parameter choice left Comparison between B factors deposited for C atoms in 1T3R and a STeM model with default weights and 50 computed modes The correlation coefficient is 0 66 right We increase th
33. o be used by MAVEN please consult the Electron Microscopy Databank at www emdatabank org Page 5 of 30 MAVEN 1 1 Section 2 1 Structure File Prepare Files elec Select Base File 7 Section 2 3 PO Im ee Sar Mae r ame aunch Atom Selector atom names A Section 2 2 wv Reset Color N gt lt Display help files in the Status Box Figure 2 0 Prepare Files Module Overview The figure above displays the Prepare Files module annotated by which section further describes each part From here users can select atoms by type perform spatial coarse graining convert density maps into PDB files and more Help files are accessible within MAVEN for each of the methods described here map2pdb Section 2 3 Status is displayed here Status Box Displays Updates dl 2 1 Selecting a Structure file Within the Prepare Files module you should first select a file PDB or MRC density map to work with If a 4 character PDB structure ID is typed in the text box then MAVEN will contact the PDB and retrieve the file In this manual this file will be referred to as the structure file or occasionally the base file It may be a raw PDB file or one that has been processed by other programs As long as the PDB format is maintained the file will be read properly The first item to note is that PDB files contain headers that give a lot of information about the expe
34. ome of them are listed here PyMOL wiki http pymolwik1 org index php Main_Page PyMOL script library http pymolwiki org index php Category Script_Library http pldserver1 biochem queensu ca rlc work pymol PyMOL community _ http sourceforge net mailarchive forum php forum_name pymol users VMD community http www ks uiuc edu Research vmd mailing_list vmd l As these software and resources already exist MAVEN does not need to reproduce them but should work seamlessly with them Many scripts plug ins and functions in VMD for trajectory analysis will work with mode animations from MAVEN Animation functions in most molecular viewers will interpret the multi MODEL PDB files as frames in an animation Thus MAVEN integrates into the already powerful analysis and visual options within these systems 5 3 Further examples of visualizations of MAVEN output in PYMOL Example 1 left the negative direction of mode 1 generated from distance dependent springs with a square dependence and PDB IT3R center The unaltered structure right the positive direction of the mode and left negative direction Top down view Generated with cartoon putty Page 28 of 30 MAVEN 1 1 Color Spectrum B Factors CA Using create to make objects for individual frames Using grid_mode and grid_slot to show the created objects next to each other Example 2 We show the initial structure same ENM model as mode 1 with mode 1
35. pe and corresponding parameters K Ca JA In iti al E N M Figure 4 5 2 Comparing one ENM to another New ENM Type In this case ile AHM_STeM ANM_ mixecdeg ANM nearestAtom Ponoi DEAR we are comparing ANMs of two different types We compare a cutoff based ANM Overlap Cumulative ekia using a cutoff of 12A with a distance _ Ig _ dependent model and a distance 05 05 dependence of 2 In both cases we use 5 the structure 1T3R and 100 modes are 13579 6 1020 calculated Again we do not find a large _ New Modes New Modes difference in the patter of fluctuations but the extent of overlap between modes from each model has decreased The distance LL z dependent springs do offer an improved agreement which is indicated by an g improvement in the B factor comparison O 50 100 150 200 o RecidoeiAtoniader the correlation increases from 0 57 to Corr Exp Initial 0 57 Erpenmeni 0 65 Corr Exp New 0 65 Initial EMM Corr Initial New 0 90 New ENN Make The Comparison Another ENM with Different Parameter 2400 poWwerANhd a Thi O Between Two Farts of the Structure a N E N M Page 22 of 30 MAVEN 1 1 4 6 Overlap Matrices Correlated motion within the structure Generate Overlap Matrices Bf Com p ute monroe using the Define subsets 1 99 100 198 Parts current of the input lt o pExamples parameters stru ctu re cha
36. pendent springs model with a power dependence Overlap Cumulative Overlap of 2 using 1T3R with a linear power 1 3 1 dependence in both cases 100 modes 0 5 a 0 5 are calculated We find that the mobility 20 of each atom is not significantly changed 0 0 13579 3 6 1020 for this structure the MSF curves are _ New Modes New Modes almost superimposed but that there has ee been a shuffling of the first 6 modes 3 a0 Overlap matrix Pearson s correlation 2 with the experimental values recorded in 10 a the PDB file are also displayed p 0 50 100 150 200 Residue Atom Index Corr Exp Initial 0 65 Espen orr Exp New 0 60 Initial ENM Corr Initial New 0 96 New ENM Make The Comparison Another EMM Vvith Different Paramete a 1 100 power ANN Thi C Between Two Farts of the Structure oe N E N M Changing the parameters used can have a large effect on the model but using a different ENM type may be a more interesting comparison The Motion Correlation Module also has the ability to perform this type of comparison In Figure 4 5 2 we compare a cutoff based ANM to a distance dependent springs model The initial model is the ENM model what is generated from the main MAVEN interface The parameters and ENM type for the new model to which the initial is compared are specified within the Motion Correlation Module see Figure 4 5 2 Page 21 of 30 MAVEN 1 1 Select an ENM ty
37. riment that derived the given structure The most critical information for MAVEN are the x y z coordinates of all atoms in the structure Commonly a model based upon the alpha carbons of proteins see CA model in section 2 2 and or 3 1 or the phosphate and or O4 atoms of RNA and DNA backbones is used Electron density maps can be imported into MAVEN by converting regions of density into coordinate points and saving those points as a PDB formatted file see map2pdb in section 2 5 Page 6 of 30 MAVEN 1 1 The original file will not be modified by any of these scripts New files will be placed in the same directory as the original structure file base file These processed files may be used for model generation via Select PDB File in the initial MAVEN window see section 1 3 Note that the analysis section 4 and export section 5 modules assume that all relevant files such as ANISOU files are in the same directory as the structure file This is the default behavior of the Prepare Files module Segment IDs are not part of the official PDB file format but are supported by MAVEN The official PDB file format does not accommodate structures with more than 99999 atoms Molecular viewers such as VMD and PyMOL as well as many other programs have added the Segment ID columns 73 76 to overcome this limitation Each segment can include up to 99999 atoms 2 2 Using the atom type selector Launching the Atom Selector will read in the str
38. s With the ever growing numbers of experimentally solved structures homology modeling and molecular simulation gathering a picture of the functional ensemble of conformations of a biomolecule is increasingly a tractable problem All three methods can either store the conformers as separate PDB files or as one multi MODEL file Both types of storage are supported by MAVEN C D E Figure 4 7 1 Explanation of PC construction A When this button is pressed the plot type that is selected will be generated and displayed B Select a plot type The options include the variance in each PC the dot product between modes and PCs Overlap the Cumulative Overlap Root Mean Squared Inner Product the percent of the PC spanning space that is shared by the normal modes and PC plots The PC plots make a scatter plot where each structure in the ensemble is a point C Pressing this button will open a file select dialog The user should select all PDB files that they wish to include in the structural ensemble from which the PCs will be constructed These may be PDB files for one conformation or multi MODEL files One could even save frames from an MD trajectory in PDB format and use them here D After processing the files and finding them to contain the same number of points as the current ENM model the number of conformers is displayed E This button helps the user to annotate PC plots Pressing it will turn the cursor into a selection crosshair W
39. sentation 3 2 6 Mixed Resolution see 2 4 for making the input PDB file 4 Analysis 4 1 B factors and their correlation with computed fluctuations 4 2 Calculating Anisotropy of motion and comparing to PDB ANISOU records MAVEN 1 1 4 3 Fluctuations in internal distances 4 4 Correlated Motion between two parts of the structure 4 5 Compare one ENM to another 4 6 Overlap Matrices Correlated motion within the structure 4 7 Compare modes to Principal Components of a structural ensemble 5 Exporting Data 5 1 Date to text file ENM summary B factors eigenvalues and mode shapes 5 2 Molecular Viewers how to visualize and animate the modes using PYMOL or VMD 5 3 Visualization Examples 6 References Please also see the MAVEN publication in BMC Bioinformatics Page 2 of 30 MAVEN 1 1 1 Overview The structures of biomolecules have become increasingly important for new biological discoveries Tens of thousands of small to medium sized protein structures have been solved experimentally as well as a growing number of polynucleotide systems and some larger entities New methods for structure determination of a wide range of biomolecules are under consideration across the world A revolution in biology occurred and is still being felt due to the availability of molecular structures and structure determinations This represents a true deepening of knowledge for biology This was due to researchers being able to explore mechani
40. sms active sites and other topological properties that were otherwise inaccessible Now that many static structures are available much attention is focused on determining their dynamics Elastic Network Modeling has become a popular method in biophysics and molecular biology for determining principal dynamics of biomolecules It is a type of normal mode analysis where the structure is typically but not necessarily treated in a coarse grained way and a simple energy function is employed This may be as simple as the one parameter Gaussian Network Model GNM or more complex methods as in the Spring Tensor model where three and four body interactions bond and torsion angles respectively are considered Here we present MAVEN a platform for generating and analyzing ENMs 1 1 What is MAVEN MAVEN is a freely available standalone application for generating and analyzing Elastic Network Models ENMs It has been developed with the goal of bringing ENMs to a wider audience and easing the steps of model generation and analysis MAVEN is freely available for download at http maven sourceforge net 1 2 Major Features A major feature of this platform is the ability to construct many types of ENMs whereas other servers and applications available are restricted to fewer types These include the standard cutoff based models distance dependent springs nearest neighbor Spring Tensor 11 and mixed resolution The nearest neighbor method generat
41. so about the overall fold For RNA or DNA systems the counterpart is a model based on the backbone phosphate atoms PDB atom type P Atoms from the sugar ring most often 04 and occasionally C2 ought to be included so that the density of points is more similar to alpha carbon models of proteins if both are included in a structure Thus nucleotide systems can be represented likewise in a coarse grained way to preserve the backbone topology Please refer to section 3 1 for making these types of models quickly by bypassing the Prepare Files module 2 3 2 Generate centroid positions United Atoms It may be of use to choose model points that do not correspond exactly to atomic points For this reason we provide the capability to compute the centroid of a group of atoms Possible selections include the centroid of each residue amino acid side chain or of each nucleotide sugar and base Note that running this script on mixed DNA RNA protein systems may result in some inappropriate centroid atoms For example asking for the ribose centroid of a protein is not meaningful If MAVEN generates a cg point Page 8 of 30 MAVEN 1 1 the resulting centroid will not be for a ribose sugar ring excluding the case of glycoproteins Similarly Nucleotide bases are not normally thought of as having side chains The following atom name lists are used for determining which group an atom is 1n nucleotide sugar Cl co nucleotid
42. tion Next we consider eight subsections of the structure The HIV 1 Protease is a homo dimer the two monomers are referred to as chain A and chain B where each monomer contains a flap domain that covers the substrate A shoulder attaches this flap to the bulk of the structure core The N Terminus can also be considered as a unit We compute overlap matrices for all unique combinations of these eight four per monomer subsets of structure in the following Parts 1 5 44 58 36 43 59 62 12 35 44 58 63 93 100 104 143 157 135 142 158 161 111 134 143 157 162 192 Names TermA FlapA shoulderA CoreA TermB FlapB shoulderB CoreB Figure 4 6 2 Heatmaps describing the details of motion direction within the first mode of motion With a larger number of subsections of the structure defined above we can better appreciate the dynamics of the structure Two example overlap matrices are shown on the left From these matrices we see that atoms in the flaps move in strong positive correlations with each other while flap A and shoulder B are almost totally anti correlated For each mode images of this type will be made for all unique pairwise combinations of the Parts Matrices which summarize all the overlap matrices for each mode are shoulderA also generated The first type visible to the left has the aa fraction of each overlap matrix that is above the thr
43. to be used but Perl and C code can be executed with appropriate compilers We do not recommend using MAVEN on a computer with less than 1GB of RAM unless only small models will be generated lt 500 points As the size of systems increases MAVEN will require more memory It is recommended to have at least 2 GB of RAM and a processor that operates at or above 2 5 GHz but analysis of some structures may require more If you run into memory barriers most of the ENM functions have an optional parameter that is the number of requested modes Requesting a smaller number of modes will decrease the required memory and computing time Disk Space The MCR requires approximately 400 MB of disk space for use MAVEN requires only 1 5 MB of space Documentation and examples total at worst 25 MB MAVEN depends on the MATLAB Component Runtime MCR a compiled library that interprets compiled MATLAB code It is likely that the MCR system requirements are similar to the MATLAB system requirements other than hard drive space see below which can be viewed for all operating systems and architectures at http www mathworks com products matlab requirements html 2 Model Generation Prepare Files The Prepare Files module will help you to take a raw PDB formatted file and prepare it for computations To learn more about PDB and the PDB file format please see www rcsb org For information regarding the MRC electron density maps that are able t
44. ucture file list all atom types in the structure file and initialize a simple molecular viewer this may take a few moments Atom types can be selected by clicking on their names in the list lower right of the Prepare Files module and then pressing the Select button You may select as many types as you wish by holding the shift and ctrl keys while clicking Selected atom types are colored yellow in the viewer Clicking the Save button will make a PDB file containing only the selected atom types all atoms that have been colored yellow To clear the atoms currently selected use the Reset Color button Pld SR ne ect TA Codie rene H CE grag errs Figure 2 2 Example Neder Vea 0ean Odds kvcees 8 08 aD of selecting by atom wi gasbe cr erxv GD w SVK GaaHke CLEVY GO B type To the left we show the PDB file 1T3R displayed in the Atom Selector This view does not look exactly like the other images of this protease structure as all ae SNES solvent atoms are atom types have due Base Protein Side cae a Mtoe cAP OF Mode N some sie ares RMANMA SunarRase RNAINMA SunarRase x been selected After selecting You selected 1051 atoms atom names N CA C and O protein backbone atoms and pressing the select button the corresponding Reset Color Save atoms are displayed in yellow and we are informed that 1051 atoms are selected Pressing the Save button would make a PDB file consisting of on
45. will be updated to show the selected file and full path If a 4 character PDB structure ID is typed in the text box then MAVEN will contact the PDB and retrieve the file B A list of implemented ENM types powerANM is currently selected This is the model type that implements distance dependent springs By clicking the button that is visible above a help file explaining the currently selected ENM type as well as its parameters is displayed C Parameter input For this model this parameter sets the distance dependence Thus points 7 and j will be connected by springs of strength dij D This button tells MAVEN to run the ENM defined by 1 the selected file which contains the points to use 2 ENM type and 3 its parameters 3 1 Quick ENM buttons Section 1 3 shows the MAVEN interface and points out two quick ENM buttons which are highlighted to the left with a red circle The buttons Q and Qpoa combine the Parse CA Model and Parse PO4 Model that are within the Prepare Files module with the Run This Model button on the main interface The result is that one can select an unprocessed PDB file press the Qa Qpo button and compute a standard alpha carbon phosphate O4 and alpha carbon model Thus these two buttons facilitate the ease with which these models can be computed for proteins alone or for proteins together with DNA or RNA One should exercise some caution when using this quick feature on d
46. work for any MRC volume or density file Thus any type of tomographic reconstruction or image stack that is saved in the MRC file format can be analyzed using MAVEN Page 9 of 30 MAVEN 1 1 EMDB Entry EMD 1800 Contact EMDB EMDB Home Single particle cryo electron microscopy Title analysis reveals the HIV 1 spike as a tripod structure _ Wu S R Loving R Lindqvist B Hebert H Authors Koeck P SjobergM Garoff H Source CD4 bound HIV 1 Eny Aggregation ma singleParticle resolution 21 Angstroms Map Information Map Data Image stored as Reals mode 2 Type SE Map Extent 252 00 A 252 00 A 252 00 A Sample Processing Map Information a Rr p s d Voxel spacing 3 50 3 50 3 50 A Map Statistics Minimum Maximum Average Standard es eel TF 3 5906 5 9747 0 0786 1 0262 Recommended 2 3 Contour Level Leve Map Download MAP CCP4 format emd 1800 map gz Map size 1495 Kb Figure 2 3 31 EMDB entry 1800 Map Information From this panel we can see the recommended density contour as well as the grid spacing in the density map This information is all the user needs to provide to convert the map into PDB file format The origin determines the coordinates of the first data point in voxel spacing 2 A Pre Da o A Ai E D _ o oceng A a sd e S e ee vo ei 0 Gg aN e o Ce 6 dee Q o o o r e Od bs o 2 al Figure 2 3 3i

Additional file 1

Contents

Download Pdf Manuals

Related Search

Related Contents