Home

VOTCA manual

1. 4 2 Post processing of the 1 Alternativen sebum dom nep Oa eue ER eb tecto f te eS Boltzmann Inversion 5 Generating exclusion lists s eo no em RR Ee De Statistical analysis ss lt e cs dk Ae ee ee ee BE 5 2 1 Distribution functions and tabulated potentials 52 2 Correlation analysis 24 nd 644654444000 04 Roy eed X R Force matching Gl Program input gaan oe Bowel e wx Ra 9 9 RR ee EASES 0 2 Program Outputs o s eg ad eee ee ba ee wor ox Gee we E YR 6 3 Integration and extrapolation of force files Iterative methods TA Iterative workflow control s sos moe acs Oe ee ee RO e Ask Preparing the rum a itas oho 9 Y oec hed 7 1 2 Starting the iterative process 71 3 Restarting and continuing s eosi sse 9 o ee 1 2 Tterative Boltzmann Inversioh 2x om n db m RERO wh Rh RU UR 24 Input preparation s s soa or ax dioi A re vw XR AG 1 9 Inverse Monte Carlo a cub E em Be DAE e eoe i ii 15 15 15 18 19 19 20 20 21 23 23 24 24 7 4 1 ESPResSo 8 1 Running IBI with ESPResSo General considerations 7 3 2 Additional mapping for statistics 7 3 3 Correlation groups 7 4 Pressure correction Simple pressure correction
2. P Y y y used for coarse graining 6 CHAPTER 2 THEORETICAL BACKGROUND 2 3 Iterative methods Iterative workflow control is essential for the IBI and IMC methods The general idea of iterative workflow is sketched in fig 2 2 run starts with an initial guess during the global initialization phase This guess is used for the first sampling step followed by an update of the potential The initial guess i update itself often requires additional postprocessing such as smoothing sampling interpolation extrapolation or fitting Different methods are available to update the potential for instance Iterative Boltzmann Inversion see next po section 2 4 or Inverse Monte Carlo see section 2 5 The whole procedure sena use is then iterated until a convergence criterion is satisfied 2 4 Iterative Boltzmann Inversion Iterative Boltzmann inversion IBI is a natural extension of the Boltzmann inversion method Since the goal of the coarse grained model is to reproduce the distribution functions of the reference system as accurately as possible one can also iteratively refine the coarse grained potentials using some nu merical scheme In IBI the potential update AU is given by 6 Figure 2 2 Block scheme of an itera U n0 Ui 2 15 tive method pm ref AUC 2 16 ref Here A 0 1 is a numerical factor which helps to stabilize the scheme The co
3. close Sout where the first line opens the file for output to a gzipped file and the blockfile is generated by appending information blocks The next to last line contains a special TCL variable that contains the list of particles to be taken into account during the RDF calculation The blockfile itself can be ordered in any way and can contain as much information as the user needs The script above represents the minimal amount of information that has to be supplied to voTCA For examples on generated blockfiles and on scripts to generate such blockfiles see the Tutorials package tutorials methanol ibm espresso conf esp gz tutorials methanol ibm espresso generate esp from gro 1 more information on ESPResSo blockfiles see the ESPResSo user guide 35 36 CHAPTER 8 ESPRESSO INTERFACE tutorials propane ibm espresso conf esp gz tutorials propane ibm espresso generate esp from gro The XML settings file contains several pieces of information specific to ESPResSo entries that are common with GROMACS are not described here cg non bonded inverse espresso lt 1 gt provides the name of the TCL vari able containing the list of typel particle IDs involved in the typel type2 RDF calculation cg non bonded inverse espresso index2 same as previously for the list of type2 particle IDs cg inverse program should be espresso cg inverse
4. espresso bin the name or path of the executable e g Espresso bin equi snapshots trash so many snapshots before analyzing the data table bins bin size for table table end distance cutoff lt blockfile gt input blockfile containing all simulation parameters gzipped format n steps number of MD steps to integrate between each snapshot n snapshots number of snaphsots before RDF calculation See the Tutorials package for XML settings file examples tutorials methanol ibm espresso settings xml tutorials propane ibm espresso settings xml Chapter 9 Advanced topics 9 1 Customization Each sub step of an iteration and all direct calls can be adjusted to the user needs The internal part of the iterative framework is organized as follows all scripts are called using two keywords csg call keyl key2 For example csg call update imc calls the update script for the inverse Monte Carlo pro cedure The corresponding keywords are listed in sec 10 4 or can be output directly by calling csg call list It is advised not to change already implemented scripts To customize a script or add a new one copy the script to your own directory set by inverse scriptdir and redirect its call by creating your own csg table file in this directory which looks like this keyl key2 scriptl options key3 key4 script2 If the local keys are already in use the existing call will be overloaded As an example we will ill
5. ESSE o goo b Rok strio 61 10 4 40 postadd plot sh lt xg ERR RSS 61 T0 4 41postupd pressuresh 4 2n oe Ro RR mom xke xn 62 10 4 42 postupd scal sh g s a Rok mw Oe mo A 62 10 4 43 postupd smooths omo Re RR Exo mom koe Rs 62 10 4 44postupd_splinesmoothsh 0 05 ee eee eee 62 10 4 45 potential_to espressO sh 2 ee x es 62 10 4 46 potential to Eromacs sli issu xs e ea RR RU hee 63 10 4 47 genericsh a s iaw somos A e a o o be eee 63 10 4 48 prepare __generic_espresso sh 63 10 4 49 prepare generic gromaessh i v mem y ec RR os 63 10 4 50prepare generic singlesh bee xk x RR 64 10 4 51 prepare 3buksh se a deca exo mo mor 9 Rem Rome c m om Yo o ex ROO 64 104L82prepare Jus gpa Se GER eee ead 64 10 4 53pressure cor simplepl 2c re eret 64 10 4 54 pressure eor wjlepl lt uo be Ex ex ERR uu 64 10 45 55 resamplo target ab os ey guod bebe G ETE e ee Um e 65 0 4 56 run espresso i 2o oos west Em eee ode d e eee gt 65 Pun BrOmecs sb erred IOS Pe e Re 65 10 4 58 80lye eresi sre E UR S 66 10 2 5980lv mumpysli ee he uere x den hw 66 10 4 6080lve octavesh e 22 ees be Rm o RERO ERR a Rd 66 10 4 01 table compare ace me OR
6. Generate options file iteratively refined Start iterations csg inverse lt options xml gt Monitor first couple of iterations Check output Many parameters can be tuned on the fly fi Figure 7 1 Flowchart to perform iterative Boltzmann inversion 7 1 Iterative workflow control Iterative workflow control is essential for the IBI and IMC methods The general idea of iterative workflow is sketched in fig 7 2 During the global initialization the initial guess for the coarse grained potential is calculated from the reference function or converted from a given potential guess into the internal format The actual iterative step starts with an iteration initialization It searches for possible checkpoints and copies and converts files from the previous step and the base directory Then the simulation run is prepared by converting potentials into the format required by the external sampling program and the actual sampling is performed After sampling the phasespace the potential update is calculated Often the update requires postprocessing such as smoothing interpolation extrapolation or fitting to an analytical form Finally the new potential is determined and postprocessed If the iterative process continues the next iterative step will start to initialize 25 26 CHAPTER 7 ITERATIVE METHODS Initialize global variables paths to Scripts executables and user defined scripts Convert target distribution functio
7. bins arg 50 bins out arg Output file rmax arg rmax default for r min of all box vectors 2 else 1 scale arg 1 scale factor for the density molname arg molname filter arg filter bead names ref arg reference zero point Trajectory options tr j arg atomistic trajectory file begin arg 0 skip frames before this time first frame arg 0 start with this frame nframes arg process the given number of frames 10 1 4 csg dump Print atoms that are read from topology file to help debugging atom naming Allowed options h help produce this help message top arg atomistic topology file Mapping options cg arg OPTIONAL coarse graining mapping definitions xml file If no file is given program acts on original trajectory Specific options 1 display exclusion list instead of molecule list 10 1 5 csg fmatch Perform force matching also called multiscale coarse graining 10 1 PROGRAMS 41 Allowed options h help produce this help message top arg atomistic topology file options arg options file for coarse graining trj force arg coarse grained trajectory containing forces of already known interac tions Mapping options cg arg coarse graining mapping definitions xml file map ignore arg list of molecules to ignore separated by no map disable mapping and act on original trajectory Trajectory options tr j arg atomistic trajectory file begin arg
8. 0 skip frames before this time first frame arg 0 start with this frame nframes arg process the given number of frames 10 1 6 csg gmxtopol Create skeleton for gromacs topology based on atomistic topology and a mapping file File still needs to be modified by the user Allowed options h help produce this help message top arg atomistic topology file out arg output topology will create top and in future also itp Mapping options cg arg coarse graining mapping definitions xml file map ignore arg list of molecules to ignore separated by no map disable mapping and act on original trajectory 10 1 7 csg imcrepack This program is internally called by inversion scripts to kick out zero entries in matrix for inverse Monte Carlo It also extracts the single potential updates out of the full solution Allowed options in arg files to read out arg files to write unpack arg extract all tables from this file help display help message 10 1 8 csg inverse Start the script to run ibi imc etc or clean out current dir Usage csg inverse OPTIONS options settings xml clean Allowed options h help show this help N do iterations N only do N iterations wall time SEK Set wall clock time options FILE Specify the options xml file to use 42 CHAPTER 10 REFERENCE debug enable debug mode with a lot of information nocolor disable colors Examples csg inverse options cg
9. 7 4 2 Advanced pressure correction 7 4 3 Runtime optimization 7 5 Thermodynamic force interface Advanced topics 9 1 Customization 9 2 Used external packages GroMaCs s i s at oed gos ESPRESSO hen pte ke eS Gnuplot 22e o9 ew 9 2 1 9 2 2 9 2 3 9 2 4 9 2 5 9 2 6 10 Reference 10 1 Programs csg boltzmann CSS 20s s as aunt Bk peces OSE density s eaea ge dump sss oe mde ese os a ee xo d 10 1 1 10 1 2 10 1 3 10 1 4 10 1 5 10 1 6 10 1 7 10 1 8 10 1 9 10 1 11 10 1 12 10 1 13csg stat GNU Octave Matlab uu HR Ros xD csg gmxtopol csg imcrepack dint BAA Sob obe ees CSE AMD zu s sont DR ee es 10 1 10csg part dist csg property csg resample 10 1 14multi g rdf 10 2 Mapping file 10 3 Settings file 10 3 1 Interaction options 10 4 Scripts 10 4 1 10 4 2 10 4 3 10 4 4 10 4 5 10 4 6 10 4 7 10 4 8 10 4 9 RDF to POT pl add POT pl add pot generic sh apply prefactor pl calc density gromacs sh cale pressure espresso sh calc pressure gromacs sh calc rdf espresso sh calc rdf generic sh 10 4 10 cale_ thermforce sh 10 4 11 configuration compare py 10 4 12convergence check default sh CONTENTS CONTENTS 10 4 13density symmetrize py sae a moe e e Ea rns 55 T0 T4 Cioppl uoo uro C aR ewe a REC ee 55 shiit c pto mox ot p e a ee uh Een ue 55
10. For example 1 AA bond 10 refers to the 10th bond named AA bond in molecule 1 To specify a couple of interactions during analysis either give the interactions separated by a space or use wildcards e g To exit the interactive mode use the command q If analysis commands are to be read from a file use the pipe or stdin redirects from the shell cat commands csg boltzmann topol top trj traj trr cg mapping xml 5 2 1 Distribution functions and tabulated potentials Distribution functions tabulated potentials can be created with the hist tab command For instance to write out the distribution function for all interactions of group AA bond where AA bond is the name specified in the mapping scheme to the file AA txt type hist AA txt The command hist set 5 2 STATISTICAL ANALYSIS 21 prints a list of all parameters that can be changed for the histogram the number n of bins for the table bounds min and max for table values scaling and normalizing a flag periodic to ensure periodic values in the table and an auto flag If auto is set to 1 bounds are calculated automatically otherwise they can be specified by min and max Larger values in the table might extend those bounds specified by parameter extend To directly write the Boltzmann inverted potential the tab command can be used Its usage and options are very similar to the hist command If tabulated potentials are written sp
11. The type of pressure correction is selected by setting inverse post update options pressure type 7 4 PRESSURE CORRECTION 31 7 41 Simple pressure correction In ref 6 a simple linear attractive potential was added to the coarse grained potential AV r 1 2 7 1 Tcutof f with prefactor A sgn AP 0 1kgT min 1 fAP 7 2 Ap and scaling factor f and Prarget can be specified in the settings file as inverse post update options pressure simple scale and inverse p_ target As an example for a block doing simple pressure correction every third interaction is lt post_update gt pressure lt post_update gt lt post_update_options gt lt pressure gt lt type gt simple lt type gt lt do gt 0 0 1 lt do gt lt simple gt lt scale gt 0 0003 lt scale gt lt simple gt lt oressure lt post_update_options gt Here inverse post update options pressure simple scale is the scaling factor f In order to get the correct pressure it can become necessary to tune the scaling factor f during the iterative process 7 4 2 Advanced pressure correction In 11 a pressure correction based on the virial expression of the pressure was introduced The potential term remains as in the simple form while a different sturcture of the A factor is used 9 2 Tout A ua 7 3 0 This factor requires the particle density p as additional input parameter which is adde
12. Used xml options cg inverse method bondtype inverse target max min name step 10 4 51 prepare ibm sh Informs users that ibm was renamed to ibi Usage prepare ibm sh 10 4 52 prepare imc sh This script initializes potentials for imc Usage prepare imc sh 10 4 53 pressure cor simple pl This script calls the pressure corrections dU 1 r r c where A 0 1k_ BT x max l p cur target scale sgn p cur p target Usage pressure cor simple pl p cur outfile Used xml options cg inverse kB T inverse p target inverse post update options pressure simple scale max min step 10 4 54 pressure cor wjk pl This script calls the pressure corrections like in Wan Junghans amp Kremer Euro Phys J E 28 221 2009 Basically dU A 1 r r c with A max 0 1k B T Int sign p cur p target and Int is the integral from Eq 7 in the paper Usage pressure cor wjk pl p cur outfile Used xml options cg inverse kB T inverse p target 10 4 SCRIPTS 65 inverse particle dens inverse post update options pressure wjk scale default 1 0 max min name step 10 4 55 resample target sh This script resamples target distribution to grid spacing of the setting xml file Usage resample target sh Used xml options bondtype inverse target max min name step 10 4 56 run espresso sh This script runs espresso for the Inverse Boltzmann Method Usage run espresso sh Used external packa
13. nocolor disable colors Examples inverse sh options cg xml inverse sh 6 options cg xml Used xml options cg inverse cleanlist default empty cg inverse convergence check default none cg inverse filelist default empty cg inverse iterations max cg inverse method cg inverse program cg inverse scriptdir default empty cg inverse simulation background default no 10 4 27 linsolve m This script has no help 10 4 28 linsolve octave This script has no help 10 4 29 linsolve py This script has no help 10 4 30 merge tables pl Merge two tables Usage merge tables pl OPTIONS source dest out Allowed options v version Print version h help Show this help message withflag only change entries with specific flag in src noflags don t copy flags novalues don t copy values Examples merge tables pl intable intable2 outtable 59 60 CHAPTER 10 10 4 31 post add sh This script makes all the post update Usage post add sh 10 4 32 post add single sh This script makes all the post update with backup for single pairs Usage post add single sh Used xml options inverse post add default empty name 10 4 33 post update generic sh This script makes all the post update Usage post update generic sh Used xml options cg inverse method 10 4 34 post update generic single sh This script makes all the post update with backup for single pairs incl backups
14. Additional post processing of U after dU added to potential This is a list of scripts separated by spaces which are called See section on iterative framework for details post add options Contains all options of post add scripts This section contains all options for post add scripts convergence weight weight factors for the convergence of the interaction should be a list of same length as add options convergence what default 1 what list for what to calc the convergence dist pot default dist copyback filelist list of files to copy to the main dir overwrite Contains all options of the overwrite postadd scripts do pattern for overwrite postadd script 1 do 0 do not plot fd file descriptor to use default 8 make it unique if you want to plot multiple things gnuplot bin gnuplot binary to use default gnuplot gnuplot opts extra options to give to gnuplot bin e g persist if one uses kill kill kill all processes with that name before ploting e g gnuplot x11 this is more reliable than using multiplot script plot script to give to gnuplot post update Additional post processing of dU before added to potential This is a list of scripts separated by spaces which are called See section on iterative framework for details post update options Contains all options of post update scripts pressure Contains all options of the pressure correction scripts do pattern for pressure correction 1 do 0 do not To do pre
15. Usage post update generic single sh Used xml options inverse post update default empty name 10 4 35 postadd acc convergence sh postadd accumulate convergence script accumulate name conv of all steps Usage postadd acc convergence sh infile outfile Used xml options name 10 4 36 postadd_convergence sh REFERENCE postadd convergence script calcs int of name DIST tgt name DIST new 2 and saves it to name conv DIST is dist but changed by onvergence what option usage postadd convergence sh infile outfile Used xml options inverse post add options convergence weight default 1 inverse post add options convergence what default dist max 10 4 SCRIPTS 61 min name step 10 4 37 copyback sh postadd copyback script copies files back to the maindir use name in filename as replacement for the interaction name Usage postadd copyback sh infile outfile Used xml options inverse post options copyback filelist default empty name 10 4 38 postadd dummy sh postadd dummy script does nothing useful to overwrite default by nothing Usage postadd dummy sh infile outfile 10 4 39 postadd overwrite sh postadd overwrite script overwrites potential of all other interactions with this one Usage postadd overwrite sh infile outfile Used xml options inverse post add inverse post add options overwrite do default 1 cg non bonded name name 10 4 40 postadd plot sh po
16. WARNING The ESPResSo interface only supports the Iterative Boltzmann Inver sion scheme It does not support Inverse Monte Carlo or Force Matching 8 1 Running IBI with ESPResSo While ESPResSo 12 is not capable of simulating atomistic systems it is possible to coarse grain molecules from existing radial distribution functions In addition to the target RDFs the user needs to provide two files e Blockfile e XML settings file The blockfile contains all the initial ESPResSo parameters to start the first simulation step time step box size temperature friction coefficient of the thermostat verlet skin etc It also includes the initial positions velocities particle types masses molecule IDs of all the particles Including velocities is important to start at the correct temperature Topology can be specified by including the bond descriptions between particles Interactions need also to be present as well as the thermostat itself In this respect the blockfile contains all the necessary information required to directly start the simulation from ESPResSo variable to initial structure to topology to interactions An example blockfile can easily be created by the following commands set out open gzip c conf esp gz w blockfile out write variable all blockfile out write particles list id type molecule mass pos v blockfile out write interactions blockfile out write thermostat blockfile Sout write tclvariable list 1151
17. allows the specification of spline interpolation spfit the calculation of derivatives derivative and comments comment Check the help help for further information It is important to note that the values min and max don t correspond to the minimum and maximum value in the input file but to the range of values the potential is desired to cover after extrapolation Therefore values in min max that are not covered in the file are automatically marked by a flag o for out of range for extrapolation in the next step The potential don t have to start at 0 this is done by the export script to xvg automatically Extrapolation The following line csg call table extrapolate options table resample pot table extrapolate pot calls the extrapolation procedure which processes the range of values marked by csg_ resample The input file is table resample pot created in the last step After resampling all values in the potential file that should be used as a basis for extrapolation are marked with an i while all values that need extrapolation are marked by o The command above now extrapolates all o values from the i values in the file Available options include averaging over a certain number of points avgpoints changing the functional form function default is quadratic extrapolating just the left or right region of the file region and setting the curvature curvature The output table extrapolate pot of the extrapolat
18. arg 0 skip frames before this time first frame arg 0 start with this frame nframes arg process the given number of frames 10 1 2 csg call This script calls scripts for the iterative framework Usage csg call OPTIONS keyl key2 Allowed options 1 1list Show list of all script cat Show the content of the script show Show the path to the script show share Shows the used CSGSHARE dir and exits scriptdir DIR Set the user script dir Used if no optins xml file is given Default empty simprog PROG Set the simprog Used if no options xml file is given Default empty options FILE Specify the options xml file to use log FILE Specify the log file to use Default stdout ia type type Specify the interaction type to use 39 40 CHAPTER 10 REFERENCE ia name name Specify the interaction name to use nocolor Disable colors debug Enable debug mode with a lot of information h help Show this help Examples csg call table smooth ARGUMENTS csg call show run gromacs 10 1 3 csg density Calculates the mass density distribution along a box axis or radial density profile from reference point Allowed options h help produce this help message top arg atomistic topology file Mapping options cg arg OPTIONAL coarse graining mapping definitions xml file If no file is given program acts on original trajectory Specific options axis arg r x y z r density axis r spherical
19. calculating the reference forces for L snapshots we can write down N x L equations I 1 N l 1 L 2 20 Here pus is the force on the bead I and is the coarse grained representation of this force The index enumerates snapshots picked for coarse graining By running the simulations long enough one can always ensure that M N x L In this case the set of equations 2 20 is overdetermined and can be solved in a least squares manner is in principle a non linear function of its parameters g Therefore it is useful to represent the coarse grained force field in such a way that equations 2 20 become linear func tions of g This can be done using splines to describe the functional form of the forces 9 Implementation details are discussed in ref 1 Note that an adequate sampling of the system requires a large number of snapshots L Hence the applicability of the method is often constrained by the amount of memory available To remedy the situation one can split the trajectory into blocks find the coarse grained potential for each block and then perform averaging over all blocks CHAPTER 2 THEORETICAL BACKGROUND 3 Input files 3 1 Mapping files Mapping relates atomistic and coarse grained representations of the system It is organized as follows for each molecule type a mapping file is created When used as a command option these files are combined in a list se
20. center of mass to zero comment arg store a comment in the output table help produce this help message 10 1 11 csg_ property Helper program called by inverse scripts to parse xml file Allowed options help produce this help message path arg list option values that match given criteria 10 1 PROGRAMS 43 filter arg list option values that match given criteria print arg list option values that match given criteria file arg xml file to parse short short version of output with path include path of node in output 10 1 12 csg resample Change grid and interval of any sort of table files Mainly called internally by inverse script can also be used to manually prepare input files for coarse grained simulations Allowed options help produce this help message in arg table to read out arg table to write derivative arg table to write grid arg new grid spacing min step max If grid is specified only interpolation is performed type arg akima cubic akima linear If option is not specified the default type akima is assumed fitgrid arg specify fit grid min step max If grid and fitgrid are specified a fit is performed nocut Option for fitgrid Normally values out of fitgrid boundaries are cut off If they shouldn t choose nocut comment arg store a comment in the output table boundaries arg natural periodic derivativezero sets boundary conditions 10 1 13 cs
21. fmatch Note that the topology files can have a different contents as bonded interactions are not pro vided in all formats In this case mapping files can be used to define and relabel bonds Also note that the default setting concerning mapping varies individually between the pro grams Some have a default setting that does mapping such as csg map use no map to disable mapping and some have mapping disabled by default e g csg stat use cg to enable mapping 3 5 Setting files xo lt non bonded gt lt non bonded interactions lt name gt A A lt name gt name of the interaction gt lt typel gt A lt typel gt types involved in this interaction gt lt type2 gt A lt type2 gt min 0 min lt dimension grid spacing of tables lt max gt 1 36 lt max gt lt step gt 0 01 lt step gt lt inverse gt specific commands lt inverse gt specific section for inverse boltzmann force matching etc lt non bonded gt ego Figure 3 3 Abstract of a settings xml file See sec 7 1 1 for a full version A setting file is written in the format xml It consists of a general section displayed above and a specific section depending on the program used for simulations The setting displayed above is later extended in the sections on iterative boltzmann inversion csg inverse force matching csg fmatch or statistical analysis csg stat Generally csg stat is an analysis tool w
22. gt lt kBT gt 1 6629 lt kBT gt 300 0 00831451 gromacs units gt lt program gt gromacs lt program gt use gromacs to sample gt lt gromacs gt gromacs specific options gt equi time 10 equi time lt table bins 0 002 table bins grid for tablex xvg gt pot max 1000000 pot max cut the potential at value gt table end 2 0 table end lt xtend the tables to value gt lt topol gt topol tpr lt topol gt topology trajectory files gt ignore so many frames lt these files are copied for each new run lt filelist gt grompp mdp topol top table xvg index ndx le log file gt gt iterations max 300 iterations max number of iterations method ibi method inverse Boltzmann or inverse MC gt log lt restart_file gt restart gt Figure 7 3 settings xml file specifies interactions to be refined grid spacings sampling engine and the iterative method The complete file can be found in the propane ibm tutorial 7 2 ITERATIVE BOLTZMANN INVERSION 29 for a file called done in the current directory If this file is found the program assumes that the run is finished To extend the run simply increase inverse iterations in the settings file and remove the file called done After that csg inverse can be restarted which will automatically recognize existing
23. index2 max min name step typel type2 54 CHAPTER 10 REFERENCE 10 4 9 calc rdf generic sh This script implemtents statistical analysis for the iterative Boltzmann inversion using generic csg tools csg stat Usage calc rdf generic sh Used xml options cg inverse sim prog equi time default 0 cg inverse 9sim prog first frame default 0 cg inverse gromacs rdf topol default topol tpr cg inverse gromacs traj type default xtc cg inverse program 10 4 10 calc thermforce sh This script cales the thermoforce out of gromacs density for the AdResS therm force Usage calc thermforce sh infile outfile Used xml options cg inverse gromacs mdp default grompp mdp max min name step tfcg prefactor default empty tf prefactor tfspline end tfspline start tf spline step 10 4 11 configuration compare py Usage configuration compare py options 1 conf2 Options h help show this help message and exit eps EPS tolerance for mismatch 10 4 12 convergence check default sh Calculated the sum of all convergence files and create a file stop if the sum is bigger than a given limit Usage convergence check default sh Used xml options cg inverse convergence check options limit cg inverse convergence check options name glob default conv 10 4 SCRIPTS 55 10 4 13 density symmetrize py This script symmetrizes the density around adressc for thermodynamic force iterat
24. max have to be chosen in such a way as to avoid empty bins within the grid Determining min and 23 24 CHAPTER 6 FORCE MATCHING Reference simulation Include forces in trajectory Define map ping scheme csg dump to list atoms il csg map to map Visualize reference mapped in e g VMD Verify map ping scheme Y Setup force matching options Provide correct intervals for distri butions e g by csg_boltzmann csg stat Run force matching csg_fmatch Integrate forces to get potential csg call table integrate i Figure 6 1 Flowchart to perform force matching max by using csg stat is recommended see sec 3 5 A full description of all available options can be found in sec 10 3 6 2 Program output csg fmatch produces a separate force file for each interaction specified in the CG options file option options These files have 4 columns containing distance corresponding force a table flag and the force error which is estimated via a block averaging procedure If you are working with an angle then the first column will contain the corresponding angle in radians To get table files for GROMACS integrate the forces in order to get potentials and do extrap olation and potentially smoothing afterwards Output files are not only produced at the end of the program execution but also after every successful processing of each block The user is free to have a look at the output files and decide
25. ping to map from reference system name Name of coarse grained bead type Type of coarse grained bead cg bonded The cg bonded section contains all bonded interaction of the molecule Those can be bond angle or dihedral An entry for each group of bonded interac tion can be specified e g several groups types of bonds can be specified A specific bonded interaction can be later on addressed by MOLECULE NAME NUMBER where MOLECULE is the molecule ID in the whole topology NAME the name of the inter action group and NUMBER addresses the interaction in the group 10 3 SETTINGS FILE 45 angle Definition of a group of angles beads List of triples of beads that define a bond Names specified in cg beads separated by commas name Name of the group bond Definition of a group of bonds beads List of pair of beads that define a bond Names specified in beads separated by commas name Name of the group dihedral Definition of a group of dihedrals Since the exact functional form does not matter this combines proper as well as improper dihedrals beads List of quadruples of beads that define bond Names specified in cg beads separated by commas name Name of the group 10 3 Settings file All options for the iterative script are stored in an xml file cg Head option which contains all coarse graining options bonded Section for a bonded interaction Most of the items in here are identical to items in cg bonded so they will be desc
26. potential after the add step pot number same as dpot lt number gt but for post add If a sub step fails during the iteration additional information can be found in the log file The name of the log file is specified in the steering XML file 7 1 3 Restarting and continuing The interrupted or finished iterative process can be restarted either by extending a finished run or by restarting the interrupted run When the script csg inverse is called it automatically checks 28 CHAPTER 7 ITERATIVE METHODS cg name A A name name of lt type2 gt A lt type2 gt lt max gt 1 36 lt max gt lt step gt 0 01 lt step gt lt inverse gt lt gromacs gt lt gromacs gt lt inverse gt non bonded width 7cm lt traj gt traj xtc lt traj gt lt gromacs gt table_al xvg table_bl xvg lt filelist gt lt log_file gt inverse log lt log_fil restart file restart points 1l inverse fog lt typel gt A lt typel gt types involved in this interaction target A A dist tgt target target distribution do potential 1 0 0 do potential lt update cycles gt non bonded non bonded interactions gt the interaction gt min 0 min dimension grid spacing of tables table table A A xvg table more non bonded interactions gt general options for the inverse script
27. program and the simulation starts Once the sampling has finished analysis programs generate new distri butions which are stored in lt interaction gt dist new and new potential updates stored in lt interaction gt dpot new Before adding the update to the old potential it can be processed in the post_update step For each script that is specified in the postupdate interaction dpot new is re named to lt interaction gt dpot old and stored in lt interaction gt dpot lt a number gt before the processing script is called Each processing script uses the current potential update lt interaction gt dpot cur and writes the processed update to lt interaction gt dpot new As an example a pressure correction is implemented as a postupdate script within this framework After all postupdate scripts have been called the update is added to the potential and the new potential lt interaction gt pot new is written Additional post processing of the potential can be performed in the post add step which is analogous to the post update step except for a potential instead of an update To summarize we list all standard output files for each iterative step dist new distribution functions of the current step dpot new the final potential update created by calc update dpot number for each postupdate script the dpot new is saved and new one is created pot cur the current potential used for the actual run pot new the new
28. steps and continue after the last one If the iteration was interrupted the script csg inverse might not be able to restart on its own In this case the easiest solution is to delete the last step and start again The script will then repeat the last step and continue However this method is not always practical since sampling and analysis might be time consuming and the run might have only crashed due to some inadequate post processing option avoid repeating the entire run the script csg inverse creates a file with restart points and labels already completed steps such as simulation analysis etc The file name is specified in the option inverse restart file If specific actions should be redone one can simply remove the corresponding lines from this file Note that a file done is also created in each folder for those steps which have been successfully finished 7 2 Iterative Boltzmann Inversion 7 2 1 Input preparation This section describes the usage of IBI implemented within the scripting framework described in the previous section 7 1 It is suggested to get a basic understanding of this framework before proceeding IBI so far only supports iterative refinement of non bonded interactions An outline of the workflow for performing IBI is given in fig 7 1 To specify Iterative Boltzmann Inversion as algorithm in the script add ibi in the method section of the XML setting file as shown below lt cg gt lt inverse gt lt me
29. the CG mapping in such a way that certain atoms belong to several CG beads at the same time 4 Following ref 3 we define two sets of atoms for each of the N CG beads For each site I a set of involved atoms is defined as 1 what follows we adopt notations of ref 3 4 CHAPTER 2 THEORETICAL BACKGROUND An atom i in the atomistic model is involved in a CG site J if and only if this atom provides a nonzero contribution to the sum in eq 2 6 set of specific atoms is defined as S A 0 and 0 for all J Z I 2 9 In other words atom i is specific to site J if and only if this atom is involved in site J and is not involved in the definition of any other site The CG model will generate an equilibrium distribution of momenta that is consistent with an underlying atomistic model if all the atoms are specific and if the mass of the CG site is given by 3 1 2 x 2 10 iETr Mr and If all atoms are specific and the center of mass of a bead is used for mapping then cri the condition 2 10 is automatically satisfied 2 2 Boltzmann inversion Boltzmann inversion is mostly used for bonded potentials such as bonds angles and torsions 5 Boltzmann inversion is structure based and only requires positions of atoms The idea of Boltzmann inversion stems from the fact that in a canonical ensemble independent degrees of freedom q obey the Boltzmann distribution i e P q Z e
30. the cross correlation of the coarse grained degrees of freedom Indepen dence of the coarse grained degrees of freedom is the main assumption that allows factorization of the probability distribution and the potential eq 2 14 Hence one has to carefully check whether 2 2 BOLTZMANN INVERSION 5 this assumption holds in practice T his can be done by performing coarse grained simulations and comparing cross correlations for all pairs of degrees of freedom in atomistic and coarse grained resolution e g using a two dimensional histogram analogous to a Ramachandran plot 2 2 1 Separation of bonded and non bonded degrees of freedom When coarse graining polymeric systems it is convenient to treat bonded and non bonded interac tions separately 5 In this case sampling of the atomistic system shall be performed on a special system where non bonded interactions are artificially removed so that the non bonded interac tions in the reference system do not contribute to the bonded interactions of the coarse grained model This can be done by employing exclusion lists using csg__boltzmann with the option exc1 This is described in detail in sec 5 1 excluded Figure 2 1 Example of excluded interactions Checking the linear correlation coefficient does not guarantee statistical independence of variables for example 0 if x has a symmetric probability density P x This case is often encountered in systems
31. the name of the script and SCRIPTDIR is the cus tom script directory which can be a global or a local path Now we change the last line of my run gromacs sh to critical mpirun np 8 mdrun This completes the customization Do not forget to add SCRIPTDIR to inverse scriptdir in the setting XML file see sec 10 3 You can check the new script by running csg call scriptdir SCRIPTDIR list csg call scriptdir SCRIPTDIR run run gromacs Finally do not forget to remove the license infomation and change the version number of the script 9 2 Used external packages 9 2 1 GroMaCs Get it from www gromacs org e mdrun e grompp 9 2 2 ESPResSo Get it from www espressomd org 9 2 3 Gnuplot Get it from www gnuplot info 9 2 4 GNU Octave Get it from www gnu org 9 2 5 Matlab Get it from www mathworks com 9 2 6 NumPy Get it from http numpy scipy org Chapter 10 Reference 10 1 Programs 10 1 1 csg boltzmann Performs tasks that are needed for simple boltzmann inversion in an interactive environment Allowed options h help produce this help message top arg atomistic topology file Mapping options cg arg coarse graining mapping definitions xml file map ignore arg list of molecules to ignore separated by no map disable mapping and act on original trajectory Special options excl arg write exclusion list to file Trajectory options tr j arg atomistic trajectory file begin
32. time default 0 cg inverse sim prog first frame default 0 cg inverse gromacs topol default topol tpr cg inverse gromacs traj type default xtc cg inverse program 10 4 23 initialize step generic sh This script implements the initialization for every step in a generic way Usage initialize step generic sh Used xml options cg inverse method cg inverse program name 10 4 24 initialize step generic espresso sh This script initializes an espresso simulation Usage initialize step generic espresso sh Used xml options cg inverse espresso blockfile default conf esp gz cg inverse espresso blockfile out default confout esp gz cg inverse initial configuration default laststep 10 4 25 initialize step generic gromacs sh This script implemtents the function initialize Usage initialize step generic gromacs sh Used external packages gromacs Used xml options cg inverse gromacs conf default conf gro cg inverse gromacs conf out default confout gro cg inverse initial configuration default laststep cg inverse method 10 4 26 inverse sh Start the script to run ibi imc etc or clean out current dir Usage inverse sh OPTIONS options settings xml clean Allowed options 10 4 SCRIPTS h help show this help N do iterations N only do N iterations wall time SEK Set wall clock time options FILE Specify the options xml file to use debug enable debug mode with a lot of information
33. to stop csg fmatch provided the force error is small enough 6 3 Integration and extrapolation of force files To convert forces force to potentials pot tables have to be integrated To use the built in integration command from the scripting framework execute csg call table integrate CG CG force minus CG CG pot csg call table linearop minus CG CG d CG CG d 1 0 This command calls the table integrate pl script which integrates the force and writes the potential to the pot file In general each potential contains regions which are not sampled In this case or in the case of further post processing the potential can be refined by employing resampling or extrapolating methods See sec 4 2 for further details 7 Iterative methods The following sections deal with the methods of Iterative Boltzmann Inversion IBI and Inverse Monte Carlo IMC In general IBI and IMC are both implemented within the same framework Therefore most settings and parameters of those methods are similar and thus described in a general section see sec 7 3 Further information on iterative methods follows in the next chapters in particular on the IBI and IMC methods Generate target Either from atomistic simulation or distributions experiment either by hand pr csg gmxtopol Cenerate all files to run simulation except for missing potentials Generate coarse grained topology Specify all interactions that should be
34. traj esp 10 4 20 functions gromacs sh Useful functions for gromacs get simulation setting gets a parameter 1st argument from gromacs mdp file 2nd parameter check cutoff compared current interactions cutoff vs rvdw check temp compares k_B T in xml with temp in mpd file simulation finish checks if simulation is finished checkpoint exist check if a checkpoint exists calc begin time return the max of dt frames and eqtime calc end time return dt nsteps Used external packages gromacs Used xml options cg inverse gromacs conf out default confout gro cg inverse gromacs cutoff check default yes cg inverse gromacs equi time default 0 cg inverse gromacs first frame default 0 cg inverse gromacs gmxrc default empty cg inverse gromacs mdp default grompp mdp cg inverse gromacs mdrun checkpoint default state cpt cg inverse gromacs temp check default yes cg inverse gromacs traj type default xtc cg inverse kBT max 10 4 21 imc purify sh This scripts cleans up the dpot tables for each interaction when using IMC Usage imc purify sh Used xml options cg inverse kB T inverse do potential default 1 max min name step 58 CHAPTER 10 REFERENCE 10 4 22 imc stat generic sh This script implemtents statistical analysis for the Inverse Monte Carlo Method using generic csg tools csg stat Usage imc stat generic sh Used xml options cg inverse 9sim prog equi
35. trr file to use default xtc imc general imc specific options matlab matlab bin Name or absolute path of the matlab binary numpy 10 3 SETTINGS FILE 47 numpy bin Name or absolute path of the python binary used by the numpy solver octave octave bin Name or absolute path of the octave binary solver solver for solving a linear equation system can be octave or matlab initial configuration what initial configuration to use in every step maindir step default laststep iterations max do the given number of iterations 0 inf kBT kBT 300 0 00831451 gromacs units log file write log to this file method ibi inverse boltzmann imc inverse monte carlo program simulation package to be used restart file Name of the restart file in case a step has to be resumed scriptdir directory for user scripts e g PWD simulation simulation options background tellcsg inverse that simulation was send to the backgroud de fault no tasks number of tasks 0 auto automatic detect on linux nbsearch Grid search algorithm simple N square search or grid default is grid non bonded Section for non bonded interaction Most of the items in here are identical to items in cg bonded so they will be described in the same section 10 3 1 Interaction options This section contains all interaction option which could be contained in the non bonded or bonded section in sec 10 3 bondtype Internal alias for non bonded and bonded set automatic
36. xml csg inverse 6 options cg xml 10 1 9 csg map Map a reference trajectory to a coarse grained trajectory This program can be used to map a whole trajectory or to create an initial configuration for a coarse grained run only Allowed options h help produce this help message top arg atomistic topology file out arg output file for coarse grained trajectory Mapping options cg arg coarse graining mapping definitions xml file map ignore arg list of molecules to ignore separated by no map disable mapping and act on original trajectory Trajectory options tr j arg atomistic trajectory file begin arg 0 skip frames before this time first frame arg 0 start with this frame nframes arg process the given number of frames 10 1 10 csg part dist This program reads a topology and set of trajectory ies For every binned value of a chosen coordinate it outputs the time averaged number of particles listed by particle types Allowed options top arg topology file tr j arg trajectory file grid arg output grid spacing min step max out arg output particle distribution table ptypes arg particle types to include in the analysis arg file particle types separated by space default all particle types first frame arg first frame considered for analysis last frame arg last frame considered for analysis coord arg coordinate analyzed x y or z default shift com shift
37. 10 4 10 dpot shit npl rearset ome Rm as 55 10 4 Lf cummysh o s o a nea wd e Se She ee eee Bead 55 10 4 13 functions common sh e 9 o mtm Rr REX A 55 10 4 19functions espresso sh Rom omes ono 56 10 4 20 functions gromacssh i eee ee a 57 10 4 2 mec puritysh sna Reece Cease mo whee xx mem he 57 10 4 221mc stat 4 2029 lob eee X E RR 58 10 4 23 initialize step osos non Voce ee T enm o 58 10 4 24initialize step generic espress sh cc mer 58 10 4 25initialize step generic gromacssh 58 10 5 Doanversesh elds oS S oe Eo Re eie et dE ed 58 TOC Ve usos a soos Yee denk eR E mgr iem OR RA e ded 59 10 4 28 linsolv octave Em Em REX ERE RR a 59 10 29hnsolvejDy s x no bro SOME Re hee eee 99 o8 Bd 59 l0 4 30merge bk Rm dese Re Rok dm 59 10 431 post addsh 39s be pad ee REEF SXTS 60 10 432post add singlessh i 2 2u Ghee REB EE 60 10 45 33 post arpdate genericsh ea des egere RUE esee RR dom ende 60 10 4 34post update generic singlesh llle 60 10 4 35postadd acc convergencesh 60 10 4 36 postadd convergence si iios eoe eedem oe Rp em 60 10 4 37 postadd eopybaelksl ge e deum om t Ren A 61 10438 p stadd dummysh i ss tA oko ce eem ab he eed ad 61 10 439 postadd overwritesh
38. 3730 3737 1995 6 8 F Ercolessi and J B Adams Interatomic potentials from 1st principles calculations the force matching method Europhys Lett 26 8 583 588 1994 7 9 S Izvekov and GA Voth Multiscale coarse graining of liquid state systems J Chem Phys 123 13 134105 OCT 1 2005 7 10 WG Noid JW Chu GS Ayton and GA Voth Multiscale coarse graining and structural correlations Connections to liquid state theory J Phys Chem B 111 16 4116 4127 APR 2007 7 11 H Wang C Junghans and K Kremer Comparative atomistic and coarse grained study of water What do we lose by coarse graining Eur Phys J E 28 2 221 229 2009 31 12 Hans J rg Limbach Axel Arnold Bernward A Mann and Christian Holm ESPResSo an extensible simulation package for research on soft matter systems Computer Physics Communications 174 9 704 727 May 2006 35 71
39. C1l 1 4 1 ppn H5 1 ppn H6 lt beads gt cg bead 1 more bead definitions gt cg beads cg bonded lt bonded interactions bond lt name gt bond lt name gt lt beads gt Al Bl Bl A2 lt beads gt lt bond gt lt angle gt lt name gt angle lt name gt lt beads gt Al Bl A2 lt beads gt lt angle gt lt cg_bonded gt lt topology gt lt maps gt map mapping A gt lt name gt A lt name gt lt weights gt 12 11 1 lt weights gt lt map gt more mapping definitions gt lt maps gt cg molecule end of the molecul gt Figure 3 2 An extract from the mapping file 1 of a propane molecule The complete file can be found in the propane single molecule tutorial 12 CHAPTER 3 INPUT FILES 3 4 Trajectories A trajectory is a set of frames containing coordinates velocities and forces for the beads defined in the topology VOTCA currently supports trr xtc pdb and gro trajectory formats Once the mapping file is created it is easy to convert an atomistic to a coarse grained trajectory using csg map csg map top topol tpr trj traj trr cg propane xml out cg gro The program csg also provides the option no map In this case no mapping is done and csg map works as a trajectory converter In general mapping can be enabled and disabled in most analysis tools e g in csg stat or csg
40. MACS output from g_rdf The program automatically takes care to resample the table to the correct grid spacing according to the options provided in settings xml The initial guess is normally taken as a potential of mean force and is generated by Boltzmann inversion of the corresponding distribution function It is written in step_000 lt name gt pot new If you want to manually specify the initial guess for a specific interaction write the potential table to a file called lt name gt pot in in the folder where you plan to run the iterative procedure A list of interactions to be iteratively refined has to be given in the options file As an example the setting xml file for a propane is shown in listing 7 3 For more details see the full description of all options in ref 10 3 7 1 2 Starting the iterative process After all input files have been set up the run can be started by csg_inverse options settings xml Each iteration is stored in a separate directory named step iteration step 000 isa special folder which contains the initial setup For each new iteration the files required to run the CG simulation as specified in the config file are copied to the current working directory The updated potentials are copied from the last step step n 1 interaction pot new and used as the new working potentials step n interaction pot cur After the run preparation all potentials are converted into the format of the sampling
41. Rx m Rom x E Rl 66 10 4 62table dummysh orbem 66 10 4 63table extrapolate pl amp 4a pack s ea REX Rx RO RE 67 10 4 64table get value pl 2 ws ee ee kee eR RE eee ee ee ee tt 67 10 4 65 table getsubs tipy sk omo mm xm mom y Rn 67 10 4 660table integrate pl ope wee Sle ee ce tS at a A es oe ee A 67 vi CONTENTS 10 467 table linearopgpl luo Bok eoo komm 68 30 4 68 table smoothypl gt er sse sani RR e dr NP eae eR 68 10 42 69 table smooth borders py gt s s s ete deeem Rom Rm ke kn nn 68 TO d pO table to tabapl ai noe eedem Rom och eR Bored gt 68 TO CL table to bx tere eo euh eb ea 69 10 4 721tables jackknife p eevee a 69 l0 4 tag flesh 22 2 RO EE m EUER Eb e dE RR Gd 69 10 4 74update Ibish 12 eee ee Pees Caw RR Re kk 69 104 oupdate ibi Dob pl sia km mmy egeo Rx 69 J0 4 76update ibi singlesh 2 26g ee 6s ranni E RR RS 69 Tod CC Update e anuo S Be V dese 70 LOA mos olus a Bodo ee RO oe a aa Ge ded a e de 70 r9 update MaR i uolo wh ee ee Ro eoe ote eu Lok Rp eed dod 70 10 4 80update tf singlesb o uus s t edt oe o ee t e 70 Chapter 1 Introduction Versatile Object oriented Toolkit for Coarse graining Applications or VOTCA is a package which helps to systematically coarse grain va
42. Thermodynamic force The thermodynamic force method is an iterative procedure to determine an external field that can correct for density variations This has been prooven to be usefull for multi scale simulations where all atom and coarse grained representations are simulated concurrently in one simulation The AdresS simulation scheme provides a protocol for such simulations The thermodynamic force is updated from the density profile in each simulation step as i r a Vel 7 4 POET where p r can be either a density along one of the box axis or a radial density this is specified by the adress paramater adress type in the gromacs mdp file In order to use the thermodynamic force iteration VOTCA must be used together with the adress branch of gromacs To check whether your gromacs version support this type mdrun h and look for the tabletf option A tutorial simulation set can be found in the tutorials spce tf which performs the thermodynamic force iteration for spc e water coupled to a coarse grained spc e water The method is selected by specifying lt method gt tf lt method gt in the inverse section For each interaction type additional options have to be specified in the settings xml file To specify in which region the thermodynamic force should be nonzero the min and max properties are used A smoothing function proportional to cos r is used to make the force go smoothly to zero at the region specif
43. VOTCA USER MANUAL VOHA Versatile Object oriented Toolkit for Coarse graining Applications Modular C kernel Iterative Boltzmann inversion Scripting for iterative workflow Inverse Monte Carlo Simple integration of other simulation packages Force matching January 10 2012 Version 1 2 2 a80d968f3054 Programs version fbc0d992ac0e VOTCA development team www votca org Contents Introduction Theoretical background Map pines xe xat m hub RE SOS Oum GR ede de SURE eet di Se dis rs 2 2 Boltzmann inversion lt os d loo be hha ba as 2 2 1 Separation of bonded and non bonded degrees of freedom 2 3 Iterative methods 5422 86 hee be nce s 2 4 Iterative Boltzmann Inyergloh udo en BAO e be a AERE 2 5 Inverse Monte Carlo ien e i iaa or aae mE LERRA 2 6 Force Matching 4 4 8 pom ead Whe ox Ea eh he XUI e Bad Input files 34 Mapping files s oon 44 08 wha Ra ee ae Ee ee eub ee doe do 3 2 Verification of a mapping se lt a 0 e ok Rotor bd ee Sh be dee eee 3 8 Advanced topology DA Trajectories ga ee ee ee me eA Stak qe pase 2 0 Setting x ee oe aw Ys ae SA Le 3 0 Table So Ra ee ee e ee A Preparing coarse grained runs 4 1 Generating a topology file for a coarse grained run
44. a pdb file as follows topology base snapshot pdb gt molecules clear define name mCP first 1 nbeads 52 nmols 216 gt lt molecules gt lt topology gt where lt clear gt clears all information that was present before Old versions of GROMACS did not store molecule names In order to use this feature a recent tpr file containing molecule names should always be provided For old topologies rerun GROMACS grompp to update the old topology file If molecule information is already present in the parent topology but molecules are not named properly as it is the case with old GROMACS tpr files one can rename them using lt topology base topol tpr gt lt molecules gt lt rename name PPY3 range 1 125 gt rename name C1 range 126 250 gt lt molecules gt lt topology gt Here the file topol tpr is loaded first and all molecules are renamed afterwards 3 8 ADVANCED TOPOLOGY HANDLING 11 cg molecule lt name gt ppn lt name gt molecule name in cg representation lt ident gt ppn lt ident gt lt molecule name in atomistic topology gt lt topology gt lt topology of one molecule gt lt cg_beads gt cg bead lt definition of a coarse grained bead gt lt name gt Al lt name gt lt type gt A lt type gt lt mapping gt A lt mapping gt reference to a map gt atoms belonging to this bead gt Xbeads 1 ppn
45. ally fmatch Force matching options max Maximum value of interval for distribution sampled in atomistic MD simulation One can get this number by looking at the distribution function for this interaction For non bonded interactions it s the cut off of the interaction min Minimum value of interval for distribution sampled in atomistic MD simulation One can get this number by looking at the distribution function for this interaction For non bonded interactions it s the distance to the rdf start For CG bonds and angles the variable has the similar meaning note that for angles it is specified in radians out step Grid spacing for the output grid Normally one wants to have this parameter smaller than fmatch step to have a smooth curve without additional spline interpola tion As a rule of thumb we normally use fmatch out step which is approximately 5 times smaller than fmatch step step grid spacing for the spline which represents the interaction This parameter should not be too big otherwise you might lose some features of the interaction poten tial and not too small either otherwise you will have unsampled bins which result in an ill defined equation system and NaNs in the output inverse Contains all information relevant to iterative process do potential Update cycle for the potential update 1 means update 0 don t update 1 1 0 means update 2 iterations then don t update then repeat espresso This section contains espresso specific op
46. b Used xml options cg inverse imc matlab bin default matlab 10 4 59 solve numpy sh This script solves a linear equation system from imc using numpy Usage solve numpy sh group lt outfile gt Used external packages numpy Used xml options cg inverse imc numpy bin default python 10 4 60 solve octave sh This script solves a linear equation system from imc using octave Usage solve octave sh group lt outfile gt Used external packages octave Used xml options cg inverse imc octave bin default octave 10 4 61 table compare pl This script compares two tables Usage table compare pl infilel infile2 10 4 62 table dummy sh This script creates a dummy table with grid min step max Usage table dummy sh min step max outfile 10 4 SCRIPTS 67 10 4 63 table extrapolate pl This script extrapolates a table Usage table extrapolate pl OPTIONS in out Allowed options avgpoints A average over the given number of points to extrapolate default is 3 function constant linear quadratic or exponential sasha default is quadratic no flagupdate do not update the flag of the extrapolated values region left right or leftright default is leftright curvature C curvature of the quadratic function default is 10000 makes sense only for quadratic extrapolation ignored for other cases h help Show this help message Extrapolation methods always m dy dx y i A y i x i A x i co
47. bug debug Espresso yes no exclusions Espresso stuff to exclude first frame rash the given number of frames at the beginning of trajectory meta cmd Espresso metadynamics command to call experimental meta min sampling Espresso metadynamics minimal number of sampling experimental n_ snapshots number of snapshots Total time n_steps Makefile Make file XMLS Makefile incl cginteraction xml t2t cginteraction xml tex cgoptions xml t2t CHAPTER 10 REFERENCE config t2t mapping xml t2t mapping xml tex xml2t2t sh snapshots n steps number of steps to integrate before a snapshot pressure command Espresso command to run when calculating the pressure name or absolute path or mpirun espresso rdf command Espresso command to run when calculating the rdf name or absolute path or mpirun espresso scriptdir overwrite ESPRESSO SCRIPTS from environment with this dir success File to create if Espresso simulation was successfull table bins grid for tabulated potentials table end Espresso end of table traj Name of the output Espresso trajectory file filelist these files are copied for each new run gromacs gromacs specific options conf Name of the coordinate file read by grompp default conf gro conf out Name of the original outcome coordinate written by mdrun default confout gro cutoff check check interaction cutoffs against rvdw in mdp file yes no de fault yes equi time begin analysis after this time when using gromacs first frame trash the giv
48. can be used to convert an atomistic trajectory to a coarse grained one as it is discussed in sec 3 4 use csg boltzmann one has to first define a mapping scheme This is outlined in sec 3 1 Once the mapping scheme is specified it is possible to generate an exclusion list for the proper sampling of the atomistic resolution system id csg map to map Visualize reference mapped in e g VMD Verify map ping scheme Create exclusion list csg_boltzmann excl i Figure 5 1 Flowchart demonstrating useful options of the tool 5 1 Generating exclusion lists Exclusion lists are useful when sampling from a special reference system is needed for example for polymer coarse graining with a separation of bonded and non bonded degrees of freedom To generate an exclusion list an atomistic topology without exclusions and a mapping scheme have to be prepared first Once the tpr topology and xml mapping files are ready simply run csg boltzmann top topol tpr cg mapping xml excl exclusions txt This will create a list of exclusions for all interactions that are not within a bonded interaction of the coarse grained sub bead As an example consider coarse graining of a linear chain of three beads which are only connected by bonds In this case csg boltzmann will create exclusions 19 20 CHAPTER 5 BOLTZMANN INVERSION for all non bonded interactions of atoms in the first bead with atoms of the 3rd bead as the
49. ck convergence acc convergence shift overwrite plot default shift nonbonded shift nonbonded shift bonded CHAPTER 10 REFERENCE The script table commonly used compare Scriptname tag file sh dummy sh functions common sh inverse sh prepare generic sh prepare imc sh prepare generic sh prepare generic sh prepare generic single sh prepare generic single sh prepare generic single sh initialize step generic sh initialize step generic sh initialize step generic sh prepare ibm sh update ibm sh update ibi sh update imc sh add pot generic sh add pot generic sh add pot generic sh RDF to POT pl post update generic sh post update generic sh dummy sh post update generic single sh post update generic single sh postupd scale sh postupd pressure sh postupd splinesmooth sh postupd smooth sh dpot shift nb pl postadd dummy sh tag file sh post add sh post add single sh tag file sh postadd dummy sh postadd copyback sh postadd convergence sh postadd acc convergence sh dpot shift nb pl postadd overwrite sh postadd plot sh convergence check default sh dpot shift nb pl dpot shift nb pl dpot shift bo pl 10 4 SCRIPTS resample dpot update update imcsolver solve imcsolver solve imcsolver solve imc update update calc tf pressure cor pressure cor density table table table table table table table table table table table configuration tables run pressure rdf imc s
50. csg call convert potential another method would be suit able This is integrating the force table as follows Integrate the tabl csg call table integrate force d minus pot d multiply by 1 csg call table linearop minus pot d pot d 1 O0 Chapter 5 Boltzmann Inversion Boltzmann inversion provides a potential of mean force for a given degree of freedom It is mostly used for deriving bonded interactions from canonical sampling of a single molecule in Prepare atom vacuum e g for polymer coarse graining where istic topology it is difficult to separate bonded and non bonded degrees of freedom 5 The non bonded potentials Define map csg dump to list atoms can then be obtained by using iterative methods ping scheme or force matching The main tool which can be used to cal culate histograms cross correlate coarse grained variables create exclusion lists as well as pre pare tabulated potentials for coarse grained sim ulations is csg boltzmann It parses the whole trajectory and stores all information on bonded Generate refer aa ence trajectory interactions in memory which is useful for inter active analysis For big systems however one can csg_boltzmann to get run out of memory In this case csg stat can gistributions potentials be used which however has a limited number of tasks it can perform see sec 3 5 for an example on its usage Another useful tool is csg map It
51. ctor Second Prefactor for the thermoforce will be linear interpolated with tf prefactor molname Molecule name of this gropu used in gromacs topology prefactor Prefactor for the thermoforce f prefactor Makefile Makefile XMLS Make file incl cginteraction xml t2t cgoptions xml t2t config t2t xml2t2t sh grad density spline end End of the spline used to smooth the density spline start Start of the spline used to smooth the density spline step Grid of the spline used to smooth the density typel Only for non bonded Bead type 1 of non bonded interaction type2 Only for non bonded Bead type 2 of non bonded interaction 50 10 4 Scripts Scripts are used by csg call and inverse csg call list Keyl tag dummy function csg prepare prepare prepare prepare prepare_ single prepare_ single prepare_ single initstep initstep initstep prepare update update update add_ pot add_ pot add_ pot rdf post update post update post update post update single post update single postupd postupd postupd postupd postupd postupd postupd post post postadd postadd postadd postadd postadd postadd postadd postadd convergence check dpot pot pot Key2 file dummy common master ibi imc generic tf ibi imc tf ibi imc tf ibm ibm ibi imc ibi imc tf pot ibi imc tf ibi imc scale pressure splinesmooth smooth shift dummy tag add add single tag dummy copyba
52. d as inverse particle dens in the input file 7 4 3 Runtime optimization Most time per iteration is spent on running the coarse grained system and on calculating the statistics To get a feeling on how much statistics is needed it is recommended to plot the distribution functions and check whether they are sufficiently smooth Bad statistics lead to rough potential updates which might cause the iterative refinement to fail All runs should be long enough to produce distributions rdfs of reasonable quality Often runtime can be improved by smoothing the potential updates Our experience has shown that it is better to smooth the potential update instead of the rdf or potential itself If the potential or rdf is smoothed sharp features like the first peak in SPC E water might get lost Smoothing on the delta potential works quite well since the sharp features are already present from the initial guess By applying iterations of a simple triangular smoothing AU 0 25AU _1 0 5AU 0 25AU 41 a reasonable coarse grained potential for SPC E water could be produced in less than 10 minutes Smoothing is implemented as a post_ update script and can be enabled by adding lt post_update gt smooth lt post_update gt lt post_update_options gt 32 CHAPTER 7 ITERATIVE METHODS smooth lt iterations gt 2 lt iterations gt lt smooth gt lt post_update_options gt to the inverse section of an interaction in the settings XML file 7 5
53. d visualization program e g vmd When comparing trajectories one has to be careful since vmd opens both a gro and trr file The first frame is then the gro file and the rest is taken from the t rr file The coarse grained trajectory contains only the frames of the trajectory Hence the first frame of the atomistic run has to be removed using the vmd menu 3 3 Advanced topology handling topology is completely specified by a set of beads their types and a list of bonded interactions VOTCA is able to read topologies in the GROMACS tpr format For example one can create coarse grained topology based on the mapping file and atomistic GROMACS topology using csg gmxtopol csg gmxtopol top topol tpr cg propane xml out out top In some cases however one might want to use a pdb file which does not contain all information about the atomistic topology In this case additional information can be supplied in the XML mapping file typical example is lack of a clear definition of molecules which can be a problem for simu lations with several molecules with multiple types During coarse graining the molecule type is identified by a name tag as names must be clearly identified To do this it is possible to read a topology and then modify parts of it The new XML topology can be used with the tpr option as any other topology file For example if information about a molecule is not present at all one can create one from
54. dd table comment to the head of a file Usage tag file sh input output 10 4 74 update ibi sh This script implements the function update for the Inverse Boltzmann Method Usage update ibi sh Used xml options cg inverse program 10 4 75 update ibi pot pl This script calcs dU out of two rdfs with the rules of inverse boltzmann In addition it does some magic tricks do not update if one of the two rdf is undefined Usage update ibi pot pl new rdf target rdf cur pot outfile Used xml options cg inverse kB T 10 4 76 update ibi single sh This script implemtents the function update for a single pair for the Inverse Boltzmann Method Usage update ibi single sh Used xml options inverse do potential default 1 max 70 CHAPTER 10 REFERENCE min name step 10 4 77 update ibm sh Informs users that ibm was renamed to ibi Usage update ibm sh 10 4 78 update imc sh This script implements the function update for the Inverse Monte Carlo Method Usage update imc sh Used xml options cg inverse imc solver cg inverse program cg non bonded inverse imc group 10 4 79 update tf sh This script implements the function update for the thermodynamic force interation Usage update tf sh 10 4 80 update tf single sh This script implemtents the function update of a single interaction for the thermodynamics force iteration Usage update tf single sh Used xml options cg inverse program inverse do potential default 1 ma
55. ecial care should be taken to the parameters T temperature and the scale The scale enables volume normalization as given in eq 2 13 Possible values are no no scaling bond normalize bonds and angle normalize angles To write out the tabulated potential for an angle potential at a temperature of 300K for instance type tab set T 300 tab set scale angle tab angle pot The table is then written into the file angle pot in the format described in sec 3 6 An optional correlation analysis is described in the next section After the file has been created by command tab the potential is prepared for the coarse grained run in chapter 4 5 2 2 Correlation analysis The factorization of P in eq 2 14 assumed uncorrelated quantities csg__boltzmann offers two ways to evaluate correlations of interactions One option is to use the linear correlation coefficient command cor However this is not a good measure since cor calculates the linear correlation only which might often lead to misleading results 1 An example for such a case are the two correlated random variables X U 1 1 with uniform distribution and Y X A simple calculation shows cov X Y 0 and therefore X Y cor var X var Y A better way is to create 2D histograms This can be done by specifying all values e g bond length angle dihedral value using the command vals e g vals vals txt 1 AA bond 1 1 AAA angle A This wi
56. en number of frames at the beginning of trajectory g energy g energy bin Name or absolute path of the g_ energy binary g energy opts Additional options to Gromacs g e g P 1 g energy topol Gromacs g rdf topol file to use default topol tpr g rdf g rdf bin Name or absolute path of the g_rdf binary g rdfindex Gromacs g rdf index file to use default index ndx g rdf opts Additional options for Gromacs g_ rdf e g nopbc g rdf topol Gromacs g topol file to use default topol tpr gmxrc GMXRC to source at the startup grompp grompp bin Name or absolute path of the grompp binary grompp index Gromacs grompp index file to use default index ndx grompp opts Additional options to Gromacs grompp e g maxwarn 1 grompp topol Text Gromacs toplogy file to use default topol top mdp Gromacs mdp fie to use default grompp mdp mdrun mdrun checkpoint Name of the checkpint to use in case of restarted simula tion default state cpt mdrun command Command to run mdrun name or absolute path or mpirun mdrun mdrun opts Additional options to Gromacs mdrun e g nosum pot max cut the potential at this value gromacs bug rdf rdf topol Gromacs topol file to be used for csg_ stat default topol tpr table bins grid for table xvg table end extend the tables to this value temp check check temperture against t ref in mdp file yes no default yes topol binary Gromacs topology file to use default topol tpr traj type Gromacs trajectory type xtc
57. f available options can be found in the reference section of csg fmatch command h 6 1 Program input csg fmatch needs an atomistic reference run to perform coarse graining Therefore the trajec tory file must contain forces note that there is a suitable option in the GROMACS map file otherwise fmatch will not be able to run In addition a mapping scheme has to be created which defines the coarse grained model see sec 3 At last a control file has to be created which contains all the information for coarse graining the interactions and parameters for the force matching run This file is specified by the tag options in the XML format An example might look like the following lt fmatch section lt fmatch gt Number of frames for block averaging gt frames per block 6 frames per block Constrained least squares lt constrainedLS gt false lt constrainedLS gt lt fmatch gt example for a non bonded interaction entry gt non bonded name of the interaction name CG CG name lt typel gt A lt typel gt lt type2 gt A lt type2 gt fmatch specific stuff lt fmatch gt lt min gt 0 27 lt min gt lt max gt 1 2 lt max gt lt step gt 0 02 lt step gt lt out_step gt 0 005 lt out_step gt lt fmatch gt lt non bonded gt af age Similarly to the case of spline fitting see sec 10 1 on csg_ resample the parameters min and
58. function source an extra function file CcSg banner print big banner csg calc simple calculator b show csg tables show all concatinated csg tables get command from csg tables print the name of script belonging to certain tags 1st 2nd argument source wrapper print the full name of a script belonging to two tags 1st 2nd argu ment find in csgshare find a script in csg script search path enable logging enables the logging to a certain file 1st argument or the logfile taken from the xml file get restart file print the name of the restart file to use check for obsolete xml options check xml file for obsolete options command not found handle print and error message if a command or a function was not found Used xml options cg inverse g file default inverse log cg inverse restart file default restart points log cg inverse simulation tasks default auto cg non bonded name name 10 4 19 functions espresso sh Useful functions for espresso simulation finish checks if simulation is finished 10 4 SCRIPTS 57 checkpoint_exist check if a checkpoint exists get_simulation_setting check if a checkpoint exists Used external packages espresso Used xml options cg inverse espresso blockfile out default confout esp gz cg inverse espresso scriptdir default empty cg inverse espresso success default success esp cg inverse espresso traj default top
59. g_ stat Calculate all distributions bonded and non bonded specified in options file Optionally calculates update matrix for invere Monte Carlo This program is called inside the inverse scripts Unlike csg boltzmann big systems can be treated as well as non bonded interactions can be evaluated Allowed options h help produce this help message top arg atomistic topology file Mapping options cg arg OPTIONAL coarse graining mapping definitions xml file If no file is given program acts on original trajectory Specific options options arg options file for coarse graining do imc write out Inverse Monte Carlo data write every arg write after every block of this length if blocking is set the averages are cleared after every output do blocks write output for blocking analysis Threading options nt arg 1 number of threads Trajectory options tr j arg atomistic trajectory file begin arg 0 skip frames before this time first frame arg 0 start with this frame nframes arg process the given number of frames 44 CHAPTER 10 REFERENCE 10 1 14 multi g rdf This is a multiplexed version of g_ rdf Usage multi_g_rdf OPTIONS g_rdf_options Allowed options N NN Number of tasks Default 8 b TIME Begin time Default 0 e TIME End time n FILE Name of the index file Default index ndx o FILE xvg Name of the total output file Default rdf xvg soutput FILE xvg Name of the single ou
60. ges espresso Used xml options cg inverse espresso blockfile default conf esp gz cg inverse espresso blockfile out default confout esp gz cg inverse espresso command default Espresso bin cg inverse espresso debug default no cg inverse espresso exclusions default 0 cg inverse espresso n snapshots cg inverse espresso n steps cg inverse espresso success default success esp cg inverse espresso traj default top traj esp cg inverse method cg non bonded inverse espresso index1 cg non bonded inverse espresso index2 10 4 57 run gromacs sh This script runs a gromacs simulation Usage run gromacs sh Used external packages gromacs Used xml options cg inverse gromacs conf default conf gro cg inverse gromacs conf out default confout gro cg inverse gromacs grompp bin default grompp cg inverse gromacs grompp index default index ndx 66 CHAPTER 10 REFERENCE cg inverse gromacs grompp opts default empty cg inverse gromacs grompp topol default topol top cg inverse gromacs mdp default grompp mdp cg inverse gromacs mdrun checkpoint default state cpt cg inverse gromacs mdrun command default mdrun cg inverse gromacs mdrun opts default empty cg inverse gromacs topol default topol tpr cg inverse gromacs traj type default xtc 10 4 58 solve matlab sh This script solves a linear equation system from imc using matlab Usage solve matlab sh group lt outfile gt Used external packages matla
61. hflag only change entries with specific flag in src with errors also read and calculate errors Examples table linearop pl tmp dpot cur tmp dpot new 1 0 0 0 10 4 68 table smooth pl This script smoothes a table Usage table smooth pl infile outfile 10 4 69 table smooth borders py This script smooths the border for thermodynamic force iteration Usage table smooth borders py Allowed options xstart X X where the smoothing starts xstop X X where the smoothing stops infile FILE input file outfile FILE output file 10 4 70 table to tab pl This script converts csg potential files to the tab format as read by espresso Potential is copied in the C12 column In addition it does some magic tricks shift the potential so that it is zero at the cutoff set all values to zero after the cutoff Usage table to tab pl in pot in deriv pot outfile Used xml options cg inverse espresso table bins cg inverse espresso table end 10 4 SCRIPTS 69 10 4 71 table to xvg pl This script converts csg potential files to the format Allowed options v version print version h help show this help message type XXX change the type of xvg table Default non bonded max MAX Replace all pot value bigger MAX by MAX Possible types non bonded C12 bond thermforce C12 C6 Examples table to xvg pl type bond table in table b0 xvg 10 4 72 tables jackknife pl This script has no help 10 4 73 tag file sh A
62. hich can be used for computing radial distribution functions and analysing them As an example the command csg stat top topol tpr trj traj xtc options settings xml computes the distributions of all interactions specified in settings xml and writes all tab ulated distributions as files interaction name dist new 3 6 TABLE FORMATS 13 3 6 Table formats Distribution functions potentials and forces are returned as tables and saved in a file Those tables generally have the format x y error flag where x is input quantity e g radius r angles 0 or y is the computed quantity e g a potential and error is an optional error for y The token flag can take the values i o or u In the first case i in range describes a value that lies within the data range o out of range symbolises a value out of the data range and u stands for an undefined value The token flag will be important when extrapolating the table as described in sec 4 2 14 CHAPTER 3 INPUT FILES Chapter 4 Preparing coarse grained runs Preliminary note The coarse grained run requires the molecule topology on the one hand and suitable potentials on the other In this chapter the generation of coarse grained runs is decribed next followed by a post processing of the potential If the potential is of such a form that it allows direct fitting of a functional form the section on post processing can be skipped Instead a program of choice sho
63. ied by min and max Additonally a tf section is needed for each interaction type lt non bonded gt lt name gt SOL lt name gt lt min gt 1 4 lt min gt lt max gt 3 1 lt max gt lt step gt 0 01 lt step gt sti lt spline_start gt 0 9 lt spline_start gt lt spline_end gt 3 6 lt spline_end gt lt spline_step gt 0 4 lt spline_step gt lt molname gt SOL lt molname gt lt prefactor gt 0 01382 lt prefactor gt a7 EES lt inverse gt lt target gt dens SOL xvg lt target gt ss inverse non bonded 7 5 THERMODYNAMIC FORCE 33 Usually the density profile fluctuates too much to obtain a force directly from the gradient Thus spline interpolation is used to smooth the force To specify the spline interpolation range the spline start and spline end parameters are used These can define a larger region than between min and max as it is sometimes usefull to extend the spline fit for numerical stability The pa rameter spline step sepcifies the bin width of the fit grid see csg resample for more The field molname specifies the molecule as defined in the gromacs topology used for calculating the density The prefactor E appearing in eq 7 4 is specified in the prefactor field A target density file has to be specified for each interaction type in most cases this will containt a flat density profile at the equilibrium density po 34 CHAPTER 7 ITERATIVE METHODS 8 ESPResSo interface
64. ion Usage density symmetrize py Allowed options adressc X X center of the adress zone x value infile FILE input file outfile FILE output file 10 4 14 dpot crop pl crop the potential update at poorly sampled ends Usage dpot crop pl OPTIONS file lt a gt lt b gt Allowed options h help Show this help message Examples dpot crop pl tmp dpot cur tmp dpot new 10 4 15 dpot shift bo pl This script shifts the whole potential to minimum like it is normally done for bonded potentials Usage dpot shift bo pl infile outfile 10 4 16 dpot shift nb pl This script shifts the whole potential to the last value like it is normally done for non bonded potentials Usage dpot shift nb pl infile outfile 10 4 17 dummy sh dummy script does nothing useful to overwrite default by nothing Usage dummy sh 10 4 18 functions common sh This file defines some commonly used functions msg echos a msg on the screen and send it to the logfile if logging is enabled die make the iterative frame work stopp cat external takes a two tags and shows content of the according script do external takes two tags find the according script and excute it critical executes arguments as command and calls die if not succesful check for duplicated interactions checks for duplicated interactions 56 CHAPTER 10 REFERENCE CSg get interaction property gets an interaction property from the xml file should
65. ion step can now be used for the coarse grained run If GROMACS is used as molecule dynamics package the potential has to be converted and exported to a suitable GROMACS format as described in the final step Exporting the table Finally the table is exported to The conversion procedure requires a small xml file table xml as shown below 42 POST PROCESSING OF THE POTENTIAL 17 cg inverse lt gromacs gt pot max 1e8 pot max table end 8 0 table end table bins 0 002 table bins gromacs inverse cg where table end is the GROMACS rvdw table extension and pot max is just a number slightly smaller than the upper value of single double precision The value given in table bins corresponds to the step value of csg_resample grid min step max Using the xml file above call csg call options table xml ia type non bonded convert_potential gromacs table_extrapolate pot table xvg to convert the extrapolated values in table_extrapolate pot to table xvg The file will contain the GROMACS C12 parts only which are stored in the sixth und seventh column this can be changed by adding the ia type C6 option for the fourth and fiveth column or ia type CB option for the second and third column after csg call Ensure compatibility with the GROMACS topology See the GROMACS manual for further information To obtain a bonded table run csg call ia type bonded options
66. l to espresso Usage potential to espresso sh Used xml options cg inverse espresso table bins 10 4 SCRIPTS inverse espresso table max name 10 4 46 potential to gromacs sh This script is a wrapper to convert a potential to gromacs Usage potential to gromacs sh input output Used xml options cg inverse gromacs mdp default grompp mdp cg inverse gromacs pot max default empty cg inverse gromacs table bins cg inverse gromacs table end cg inverse gromacs table end default empty cg inverse method default empty bondtype inverse gromacs table name 10 4 47 prepare generic sh This script prepares potentials in a generic way Usage prepare generic sh Used xml options cg inverse method cg inverse program 10 4 48 prepare generic espresso sh This script implements the prepare step for espresso Usage prepare generic espresso sh Used xml options cg inverse espresso blockfile default conf esp gz cg inverse espresso blockfile out default confout esp gz 10 4 49 prepare generic gromacs sh This script does the prepare step for gromacs Usage prepare generic gromacs sh Used xml options cg inverse gromacs conf default conf gro cg inverse gromacs conf out default confout gro 63 64 CHAPTER 10 REFERENCE 10 4 50 prepare generic single sh This script implements the prepares the potential in step 0 using pot in or by resampling the target distribution Usage prepare generic single sh
67. lecule in the atomistic representation In the topology section all beads are defined by spec ifying bead name Al B1 A2 type and atoms belonging to this bead in the form residue id residue name atom name For each bead a map has to be specified which is defined later in maps section Note that bead type and map can be different which might be useful in a situation when chemically different beads A1 B1 are assigned to the same bead type After defining all beads the bonded interactions of the coarse grained molecule must be specified in the cg bonded section This is done by using the identifiers of the beads in the coarse grained model Finally in the mapping section the mapping coefficients are defined This includes a weighting of the atoms in the topology section In particular the number of weights given should match the number of beads 10 CHAPTER 3 INPUT FILES 3 2 Verification of a mapping Note that the ident tag should match the molecule name in the reference system common mistake is that beads have wrong names In this case the csg dump tool can be used in order to identify the atoms which are read in from a topology file tpr This tool displays the atoms in the format residue id residue name atom name For multicomponent systems it might happen that molecules are not identified correctly The workaround for this case is described in sec 3 3 To compare coarse grained and atomistic configurations one can use a standar
68. les are described Before using the package do not forget to initalize the variables in the bash or csh tcsh source csg installation bin VOTCARC bash Source csg installation bin VOTCARC csh More details as well as several examples can be found in ref 1 Please cite this paper if you are using the package Tutorials can be found on the VOTCA homepage WWW VOTCA ORG Chapter 2 Theoretical background 2 1 Mapping The mapping is an operator that establishes a link between the atomistic and coarse grained representations of the system An atomistic system is described by specifying the values of the Cartesian coordinates and momenta T Pis Pn 2 2 of the n atoms in the system On a coarse grained level the coordinates and momenta are specified by the positions and momenta of CG sites RN R Ry 2 3 PN P Py 2 4 Note that capitalized symbols are used for the CG sites while lower case letters are used for the atomistic system The mapping operator c is defined by a matrix for each bead J and links the two descriptions Rr 5 Chri 2 5 i 1 2 Cli Pr M R M iti pj 2 6 I pity I 2 cnr I 2 for all I 1 N If an atomistic system is translated by a constant vector the corresponding coarse grained system is also translated by the same vector This implies that for all 7 Sos 1 2 7 i l In some cases it is useful to define
69. ll create a file which contains 3 columns the first being the time and the second and third being bond and angle respectively Columns 2 and 3 can either be used to generate the 2D histogram or a simpler plot of column 3 over 2 whose density of points reflect the probability Two examples for 2D histograms are shown below one for the propane molecule and one for hexane 0 180 oss F7 b nm 0 150 1 9 2 0 21 16 18 20 22 24 26 28 30 15 20 25 0 rad 9 rad 8 rad Figure 5 2 propane his togram Figure 5 3 hexane histograms before and after the coarse grained run 17 18 22 CHAPTER 5 BOLTZMANN INVERSION The two plots show the correlations between angle and bondlength for both molecules In the case of propane the two quantities are not correlated as shown by the centered distribution while correlations exist in the case of hexane Moreover it is visible from the hexane plot that the partition of the correlations has changed slightly during coarse graining The tabulated potentials created in this section can be further modified and prepared for the coarse grained run This includes fitting of a smooth functional form extrapolation and clipping of poorly sampled regions Further processing of the potential is decribed in chapter 4 6 Force matching The force matching algorithm with cubic spline basis is implemented in the csg fmatch utility A list o
70. mate the many body potential of mean force The choice of a particular method is system specific and requires a thorough consistency 2 CHAPTER 1 INTRODUCTION check It is important to keep in mind that coarse graining should be used with understanding and caution methods should be crossed checked with each other as well as with respect to the reference system The package consists of two parts a C kernel and a scripting engine The kernel is capable of processing atomistic topologies and trajectories and offers a flexible framework for reading manipulating and analyzing topologies and generated by MD SD MC sampling trajectories It is modular new file formats can be integrated without changing the existing code Currently an interface for GROMACS 2 topologies and trajectories is provided The kernel also includes various coarse graining tools for example calculations of probability distributions of bonded and non bonded interactions correlation and autocorrelation functions and updates for the coarse grained pair potential The scripting engine is used to steer the iterative procedures Here the analysis tools of the package used for sampling e g GROMACS tools can be integrated into the coarse graining workflow if needed The coarse graining workflow itself is controlled by several Extensible Markup Language XML input files which contain mapping and other options required for the workflow control In what follows these input fi
71. n the system 3 N N 1 is then the number of all pairs Ar is the grid spacing reut M V is the total volume of the system In other words in this particular case the physical meaning of S is the number of particle pairs with interparticle distances ri ra which correspond to the tabulated value of the potential Ua 2 6 Force Matching Force matching FM is another approach to evaluate corse grained potentials 8 10 In contrast to the structure based approaches its aim is not to reproduce various distribution functions but instead to match the multibody potential of mean force as close as possible with a given set of coarse grained interactions The method works as follows We first assume that the coarse grained force field and hence the forces depends on M parameters gi gm These parameters can be prefactors of analytical functions tabulated values of the interaction potentials or coefficients of splines used to describe these potentials In order to determine these parameters the reference forces on coarse grained beads are cal culated by summing up the forces on the atoms dy Y T frm 2 19 C j 84 li where the sum is over all atoms of the CG site I see sec 2 1 The dr coefficients can in principle be chosen arbitrarily provided that the condition 1 is satisfied 3 If mapping coefficients for the forces are not provided it is assumed that dr cr see also sec 3 By
72. ns Iteration initialization into internal format prepare input files copy data of the previous step Prepare input files for the external Prepare sampling sampling program o o o D 5 lt lt lt m F 2 5 Canonical ensemble sampling with Sampling molecular dynamics stochastic dy namics or Monte Carlo techniques Analysis of the run Evaluation of dis Calculate updates tribution functions potential updates 1 AUC Postprocessing Smoothing extrapolation of potential of updates updates Ad hoc pressure correction Update potentials Uo go Auc 1 Postprocessing Smoothing extrapolation of potentials of potentials Uo Evaluation of the convergence crite rion either for AU or distribution functions Check the number of itera tions lt 5 lt Finish Figure 7 2 Block scheme of the workflow control for the iterative methods The most time consuming parts are marked in red How to start The first thing to do is generate reference distribution functions These might come from exper iments or from atomistic simulations To get reasonable results out of the iterative process the reference distributions should be of good quality little noise etc VOTCA can create initial guesses for the coarse grained potentials by boltzmann inverting the distribution function If a custom initial guess for an interaction shall be used instead the table can be provided in l
73. nstant y y0 linear y ar b mx xz 0 y_ 0 a m sasha y b b x0 2y 0 m m A y 0 exponential y a x exp b a y0 exp m 10 y0 b m y 0 quadratic y C x a b m 2 C 20 b y 0 m A 10 4 64 table get value pl This script print the y value of x which is closest to X Usage table get value pl OPTIONS X infile Allowed options h help Show this help message 10 4 65 table getsubset py This script get the a subset of a table Usage table getsubset py Allowed options xstart X X x value where the subset starts xstop X X x value where the subset stops infile FILE input file outfile FILE output file 10 4 66 table integrate pl This script calculates the integral of a table Please note the force is the NEGATIVE integral of the potential use table linearop and multiply the table with 1 Usage table integrate pl OPTIONS in out Allowed options with errors calculate error with S Add entropic contribution to force 2k_ BT r kbT NUMBER use NUMBER as k_B for the entropic part h help Show this help message Examples 68 CHAPTER 10 REFERENCE table integrate pl with S kbT 2 49435 tmp force tmp dpot 10 4 67 table linearop pl This script performs a linear operation on the y values new a y old 0 Usage table linearop pl OPTIONS in out a b Allowed options h help Show this help message wit
74. nvergence is reached as soon as the distribution function PC matches the reference distribution function or in other words the potential of mean force USD converges to the reference potential of mean force IBI can be used to refine both bonded and non bonded potentials It is primarily used for simple fluids with the aim to reproduce the radial distribution function of the reference system in order to obtain non bonded interactions On the implementation side IBI has the same issues as the inverse Boltzmann method i e smoothing and extrapolation of the potential must be used 2 5 Inverse Monte Carlo Inverse Monte Carlo is an iterative scheme which additionally includes cross correlations of distributions A detailed derivation of the IMC method can be found in ref 7 The potential update AU of the IMC method is calculated by solving a set of linear equations Sa 5 Ao AU 2 17 where _ O Sa _ Aay aU ES B 6S4 Sy ni SqSy and S the histogram of a coarse grained variable of interest For example in case of coarse graining of the non bonded interactions which depend only on the distance r between particles i and j and assuming that the interaction potential is short ranged ie U rij 0 if rij gt Teut the average value of S is related to the radial distribution function g ra by N N 1 4nr2 84 BE Gra 2 18 2 6 FORCE MATCHING T where N is the number of atoms i
75. o bead types A and B with three and two hydrogens combined with the corresponding atom as shown in fig 1 1 This representation defines the mapping operator as well as the bonded coarse grained degrees of freedom such as the bond 6 and the bond angle Apart from the bonded interactions and ug beads belonging to different molecules have non bonded interactions UAA UAB The task of coarse graining is then to Figure 1 1 Three bead derive a potential energy surface u which is a function of all coarse coarse grained model of grained degrees of freedom Note that while the atomistic bond and propane angle potentials are often chosen to be simple harmonic functions the coarse grained potentials cannot be expressed in terms of simple analytic functions Instead tabulated functions are normally used The coarse graining method defines criteria according to which the potential energy surface is constructed For example for the bond b and the angle 0 Boltzmann Inversion can be used In this case a coarse grained potential will be a potential of mean force For the non bonded degrees of freedom the package provides Iterative Boltzmann Inversion IBI or Inverse Monte Carlo IMC methods In this case the radial distribution functions of the coarse grained model will match those of the atomistic model Alternatively Force Matching FM or multiscale coarse graining can be used in which case the coarse grained potential will approxi
76. only be called from inside a for all loop CSg get property get an property from the xml file trim all strips white space from beginning and the end of all args mark done mark a task 1st argument as done in the restart file is done checks if something is already do in the restart file int check checks if lst argument is a integer or calls die with error message 2nd argument num check checks if lst argument is a number or calls die with error message 2nd argument get stepname get the dir name of a certain step number 1st argument get current step dir print the directory of the current step get last step dir print the directory of the last step get main dir print the main directory get current step nr print the main directory get step nr print the number of a certain step directory 1st argument cp from main dir copy something from the main directory cp from last step copy something from the last step get number tasks get the number of possible tasks from the xml file or determine it automatically under linux get table comment get comment lines from a table and add common information which include the hgid and other information csg inverse clean clean out the main directory add to csgshare added an directory to the csg internal search directories globalize dir convert a local directory to a global one globalize file convert a local file name to a global one Source
77. package provides a collection of scripts to handle potentials They can be modified refined integrated or inter and extrapolated These scripts are the same ones as those used for iterative methods in chapter 7 Scripts are called by csg call A complete list of available scripts can be found in sec 10 4 The post processing roughly consists of the following steps see further explanations below e manually clipping poorly sampled border regions 15 16 CHAPTER 4 PREPARING COARSE GRAINED RUNS e resampling the potential in order to change the grid to the proper format csg resample e extrapolation of the potential at the borders csg call table extrapolate e exporting the table to csg call convert potential gromacs Clipping of poorly sampled regions Regions with an irregular distribution of samples should be deleted first T his is simply done by editing the pot file and by deleting those values Alternatively manually check the range where the potential still looks good and is not to noisy and set the flags in the potential file of the bad parts by hand to o for out of range Those values will later be extrapolated and overwritten Resampling Use the command csg resample in table pot out table resample pot grid min step max to resample the potential given in file table pot from min to max with a grid spacing of step steps The result is written to the file specified by out Additionally csg resample
78. parated by a semicolon g cg protein xml solvent xml Each mapping file contains a topology of the coarse grained molecule and a list of maps Topology specifies coarse grained beads and bonded interactions between them Each coarse grained bead has a name type i list of atoms which belong it and a link to a map H4 C125 C2 C3 H9 map is set of weights cr for an atom i belonging to the i bead I It is used to calculate the position of a coarse grained bead from the positions of atoms which belong A1 B1 A2 toit Note that cr will be automatically re normalized if their sum is not equal to 1 i e in the case of a center of mass mapping one can simply specify atomic masses Figure 3 1 Atom labeling and mapping A complete reference for mapping file definitions can be from an all atom to a united atom rep found in sec 10 2 resentation of a propane molecule As an example we will describe here a mapping file of a united atom model of a propane molecule chemical structure of which is shown in fig 1 1 In this coarse grained model two bead types A B and three beads A1 B1 A2 are defined as shown in fig 3 1 We will use centers of mass of the beads as coarse grained coordinates Extracts from the propane xml file of the tutorial are shown below The name tag indicates the molecule name in the coarse grained topology The ident tag must match the name of the mo
79. resso sh potential to gromacs sh table to xvg pl functions gromacs sh run espresso sh cale pressure espresso sh cale rdf espresso sh potential to espresso sh table to tab pl functions espresso sh Script calls can be overwritten by adding a line with the 3rd column changed to table in inverse scriptdir directory 52 CHAPTER 10 REFERENCE 10 4 41 RDF to POT pl This script converts rdf to pot of mean force F r k_ BT Ing r In addtion it does some magic tricks do not crash when calc log 0 extrapolate the beginning of pot the maximum to interpolate is pot max see xml bigger value will be set to that max shift the potential so that it is zero at the cutoff set all values to zero after the cutoff Usage to POT pl infile outfile Used xml options cg inverse kBT max 10 4 2 add POT pl This script adds up two potentials In addition it does some magic tricks order of infiles MATTERS if infile2 contains an undefined value it uses the value from infilel if value for infilel and infile2 are both invalid the result is also invalid Usage add POT pl infilel infile2 outfile 10 4 3 add pot generic sh This script adds up the tables Usage add pot generic sh Used xml options name 10 4 4 apply prefactor pl This script calculates the integral of a table Usage apply prefactor pl OPTIONS in out Allowed options h help Show this help message 10 4 5 calc density gromacs
80. ribed in the same section fmatch Force matching options constrainedLS boolean variable false simple least squares true constrained least squares For details see the VOTCA paper Practically both algorithms give the same results but simple least squares is faster If you are a mathematician and you think that a spline can only then be called a spline if it has continuous first and second derivatives use constrained least squares frames per block number of frames being used for block averaging Atomistic trajectory specified with trj option is divided into blocks and the force matching equations are solved separately for each block Coarse grained force field which one gets on the output is averaged over those blocks inverse general options for inverse script sim prog generic simulation program e g GROMACS options equi time begin analysis after this time first frame trash the given number of frames at the beginning of trajectory cleanlist these files are removed after each new run convergence check type of convergence check to do convergence check options options for the convergence check limit lower limit to stop name _ glob files to check for number default conv espresso blockfile Name of the original blockfile read by Espresso default conf esp gz blockfile out Name of the original outcome blockfile written by Espresso default confout esp gz command Command to run espresso name or absolute path or mpirun espresso de
81. rious systems 1 This includes deriving the coarse grained potentials assessing their quality preparing input files required for coarse grained simulations and analyzing the latter typical coarse graining workflow includes sampling of the system of interest analysis of the trajectory using a specific mapping and a coarse graining method to derive coarse grained potentials and in case of iterative methods running coarse grained simulations and iteratively refining the coarse grained potentials In most cases coarse graining requires canonical sampling of a reference high resolution sys tem In addition iterative methods require canonical sampling of the coarse grained system The sampling can be done using either molecular dynamics MD stochastic dynamics SD or Monte Carlo MC techniques The latter are implemented in many standard simulation packages Rather than implementing its own MD SD MC modules voTCA allows swift and flexible integration of existing programs in such way that sampling is performed by the program of choice At the moment an interface to GROMACS 2 simulation package is provided The rest of the analysis needed for systematic coarse graining is done using the package tools The workflow can be exemplified on coarse graining of a propane liquid A single molecule of propane contains three carbon and eight hydrogen atoms A united atom coarse grained representation of a propane molecule has three beads and tw
82. se would contribute only to the non bonded interaction potential Note that csg boltzmann will only create the exclusion list for the fist molecule in the topology To add the exclusions to the GROMACS topology of the molecule either include the file specified by the excl option into the top file as follows exclusions include exclusions txt or copy and paste the content of that file to the exclusions section of the gromacs topology file 5 2 Statistical analysis For statistical analysis csg__boltzmann provides an interactive mode To enter the interactive mode use the trj option followed by the file name of the reference trajectory csg boltzmann top topol tpr trj traj trr cg mapping xml To get help on a specific command of the interactive mode type help command for example help hist help hist set periodic Additionally use the list command for a list of available interactions Note again that csg boltzmann loads the whole trajectory and all information on bonded interactions into the memory Hence its main application should be single molecules See the introduction of this chapter for the csg stat command If a specific interaction shall be used it can be referred to by molecule interaction group index Here molecule is the molecule number in the whole topology interaction group is the name specified in the bond section of the mapping file and index is the entry in the list of interactions
83. sh This script calcs the density for gromacs for the AdResS therm force Usage calc density gromacs sh Used xml options cginverse sim prog equi time default 0 cg inverse sim_prog first frame default 0 cg inverse gromacs topol default topol tpr cg inverse gromacs traj type default xtc 10 4 SCRIPTS 53 cg inverse program name step tf molname default tf spline_ end 10 4 6 calc pressure espresso sh This script calcs the pressure for espresso and writes it to outfile Usage calc pressure espresso sh outfile Used external packages espresso Used xml options cg inverse espresso blockfile default conf esp gz cg inverse espresso pressure command default Espresso bin 10 4 7 calc pressure gromacs sh This script calcs the pressure for gromacs and writes it to outfile Usage calc pressure gromacs sh outfile Used external packages gromacs Used xml options cg inverse gromacs g energy bin default g energy cg inverse gromacs g energy opts default empty cg inverse gromacs g energy topol default topol tpr 10 4 8 calc rdf espresso sh This script calcs the rdf for espresso Usage calc rdf espresso sh Used external packages espresso Used xml options cg inverse espresso blockfile default conf esp gz cg inverse espresso first frame default 0 cg inverse espresso rdf command default Espresso bin cg inverse espresso traj default top traj esp inverse espresso index1 inverse espresso
84. ssure correction every third step specify 0 0 1 similar to inverse do update simple Contains all options of the simple pressure correction script simple scale slope of the simple pressure correction type Pressure correction algoritm can be simple or wjk wjk Contains all options of the wjk pressure correction script wjk scale extra scaling factor of pressure correction scale scale factor for the update smooth Contains all options of the smooth script iterations number of iterations for triangular smooth splinesmooth Contains all options of the spline smooth script step grid spacing for spline fit when doing spline smoothing 10 3 SETTINGS FILE 49 target target distribution e g rdf which is tried to match during iterations to match max upper bound of interval for potential table in which calculations are performed Should be set based on reference distributions min lower bound of interval for potential table in which calculations are performed Should be set based on reference distributions name Name of the interaction The name can be arbitrary but should be unique For bonded interactions this should match the name specified in the mapping file step step size of interval for potential table in which calculations are performed If step site is too small lots of statistics is needed long runs If it s too big features in the distribtuion potentials might get lost tf Contains all information relevant to thermoforce iteration cg prefa
85. stadd plot script send a certain plot script to gnuplot Usage postadd plot sh infile outfile Used external packages gnuplot Used xml options inverse post add options plot fd default 8 inverse post add options plot gnuplot bin default gnuplot inverse post options plot gnuplot opts default empty inverse post add options plot kill default empty inverse post add options plot script 62 CHAPTER 10 REFERENCE 10 4 41 postupd pressure sh This script implements the pressure update Usage postupd pressure sh infile outfile Used xml options cg inverse program inverse post update options pressure do default 1 inverse post update options pressure type default simple max min name step 10 4 42 postupd scale sh This script implements scaling of the potential update dpot Usage postupd scale sh infile outfile Used xml options inverse post update options scale default 1 0 name 10 4 43 postupd smooth sh This script implements smoothing of the potential update dpot Usage postupd smooth sh infile outfile Used xml options inverse post update options smooth iterations default 1 name 10 4 44 postupd splinesmooth sh This script implements smoothing of the potential update dpot Usage postupd splinesmooth sh infile outfile Used xml options inverse post update options splinesmooth step max min name step 10 4 45 potential to espresso sh This script is a wrapper to convert a potentia
86. t cg_bead gt lt name gt CG lt name gt lt type gt CG lt type gt lt mapping gt A lt mapping gt lt beads gt 1 SOL CG lt beads gt lt cg_bead gt lt cg_beads gt lt 7 tonoloqu lt maps gt lt map gt lt name gt A lt name gt lt weights gt 1 lt weights gt lt map gt lt maps gt lt cg_molecule gt 7 3 3 Correlation groups Unlike IBI IMC also takes cross correlations of interactions into account in order to calculate the update However it might not always be beneficial to evaluate cross correlations of all pairs of interactions By specifying inverse imc group VOTCA allows to define groups of interactions amongst which cross correlations are taken into account where inverse imc group can be any name lt non bonded gt name CG CG name lt typel gt CG lt typel gt lt type2 gt CG lt type2 gt lt imc gt lt group gt solvent lt group gt non bonded non bonded 7 4 Pressure correction The pressure of the coarse grained system usually does not match the pressure of the full atomistic system This is because iterative Boltzmann inversion only targets structural properties but not thermodynamic properties In order correct the pressure in such a way that it matches the tar get pressure inverse p target different strategies have been used based on small modifications of the potential The correction can be enable by adding pressure to the list of inverse post update scripts
87. t interaction gt pot in As already mentioned VOTCA automatically creates potential tables to run a simulation However it does not know how to run a coarse grained simulation Therefore all files needed to run a coarse grained simulation except for the potentials that are iteratively refined must be provided and added to the filelist in the settings XML file If an atomistic topology and a mapping definition are present VOTCA offers tools to assist the setup of a coarse grained topology see chapter 4 To get an overview of how input files look like it is suggested to take a look at one of the tutorials provided on WWW VOTCA ORG In what follows we describe how to set up the iterative coarse graining run the main script continue the run and add customized scripts 7 1 1 Preparing the run To start the first iteration one has to prepare the input for the sampling program This means that all files for running a coarse grained simulation must be present and described in a separate 7 1 ITERATIVE WORKFLOW CONTROL 27 XML file in our case settings xml see sec 3 5 for details An extract from this file is given below The only exception are tabulated potentials which will be created and updated by the script in the course of the iterative process The input files include target distributions initial guess optional and a list of interactions to be iteratively refined As a target distribution any table file can be given e g GRO
88. table xml convert potential gromacs table extrapolate pot table xvg It is also possible to use angle and dihedral as type as well Internally convert potential gromacs will do the following steps e Resampling of the potential from 0 or 180 for dihedrals to table end or 180 for angles and dihedrals with step size table bins This is needed for gromacs the table must start with 0 or 180 e Extrapolate the left side to 0 or 180 expontially e Extrapolate the right side to table end or 180 expontially or constant for non bonded interactions e Shift it so that the potential is zero at table end for non bonded interactions or zero at the minium for bonded interaction e Calculate the force assume periodicity for dihedral potentials e Write to the format needed by gromacs An example on non bonded interactions csg call pot shift nonbonded table pot table pot refined csg resample grid 0 3 0 05 2 in table pot refined out table pot refined csg call table extrapolate function quadratic region left table pot refined table pot refined csg_call table extrapolate function constant region right table pot refined table pot refined 18 CHAPTER 4 PREPARING COARSE GRAINED RUNS 4 3 Alternatives Additionally to the two methods described above namely a providing the MD package directly with a functional form fitted with a program of choice or b using csg_resample csg call table extrapolate and
89. tat density prepare generic initstep generic prepare generic initstep generic convert potential convert potential functions run pressure rdf convert potential convert potential functions target crop ibi single ibi pot matlab matlab octave octave numpy numpy purify tf tf single thermforce apply prefactor simple wjk symmetrize add integrate extrapolate merge smooth linearop dummy get value getsubset smooth borders compare compare jackknife gromacs gromacs gromacs gromacs gromacs gromacs gromacs espresso espresso gromacs xvg gromacs espresso espresso espresso espresso tab espresso 51 resample target sh dpot crop pl update ibi single sh update ibi pot pl solve matlab sh linsolve m solve octave sh linsolve octave solve numpy sh linsolve py imc purify sh update tf sh update tf single sh cale thermforce sh apply prefactor pl pressure cor simple pl pressure cor wjk pl density symmetrize py add POT pl table integrate pl table extrapolate pl merge tables pl table smooth pl table linearop pl table dummy sh table get value pl table getsubset py table smooth borders py table compare pl configuration compare py tables jackknife pl run gromacs sh cale pressure gromacs sh cale rdf generic sh imc stat generic sh cale density gromacs sh prepare generic gromacs sh initialize step generic gromacs sh prepare generic espresso sh initialize step generic esp
90. thod gt ibi lt method gt lt inverse gt fap 7 3 Inverse Monte Carlo In this section additional options are described to run IMC coarse graining The usage of IMC is similar to the one of IBI and understanding the use of the scripting framework described in chapter 7 1 is necessary WARNING multicomponent IMC is still experimental 7 3 1 General considerations In comparison to IBI IMC needs significantly more statistics to calculate the potential update 1 It is advisable to perform smoothing on the potential update Smoothing can be performed as described in sec 7 4 3 In addition IMC can lead to problems related to finite size for methanol an undersized system proved to lead to a linear shift in the potential 1 It is therefore always necessary to check that the system size is sufficiently large and that runlength csg smoothing iterations are well balanced 7 3 2 Additional mapping for statistics The program csg_ stat is used for evaluating the IMC matrix Although the matrix only acts on the coarse grained system here it still needs a mapping file to work This will improve with one of 30 CHAPTER 7 ITERATIVE METHODS the next releases to simplify the setup The mapping file needs to be a one to one mapping of the coarse grained system e g for coarse graining SPC E water the mapping file looks as follows cg molecule name SOL name lt ident gt SOL lt ident gt lt topology gt lt cg_beads gt l
91. tions in case espresso is used as simulation program index1 Index list of typel Name of the Tcl variable containing all index1 particles that is contained in the espresso blockfile index2 Index list of type2 Name of the Tcl variable containing all index2 particles that is contained in the espresso blockfile table Name of file for tabulated potential of this interaction This file will be 48 CHAPTER 10 REFERENCE created from the internal tabulated potential format for every run Note though that the original espresso blockfile needs to contain the name of that table as the tabulated interaction see tutorial methanol ibi espresso for details gromacs This section contains gromacs specific options in case gromacs is used as simulation program grpl Name of energy group of bead typel using in the g_ rdf index file grp2 Name of energy group of bead type2 using in the g_ rdf index file table Name of file for tabulated potential of this interaction This fill will be created from the internal tabulated potential format for every run imc Section containing inverse monte carlo specific options group Group of interaction Cross correlations of all members of group are taken into account for calculating the update If no cross correlations should be calculated interactions have to be put into different groups p target partial pressure of this species particle dens particle density of this species for wjk pressure correction post add
92. tput files Default rdf_NP xvg used trunc of name given by where NP is replaced later by the number of the process log FILE Name of logfile Default rdf NP log used of name given by o where NP is replaced later by the number of the process cmd CMD Change the gromacs command to run Default g_ rdf single Run only one task at the time debug Enable debug output q quiet Be a little bit quiet h help Show this help Examples multi g rdf e 1 multi g rdf e 1 bin 0 05 10 2 Mapping file The root node always has to be cg molecule It can contain the following keywords ident Molecule name in reference topology maps Section containing definitions of mapping schemes map Section for a mapping for 1 bead name Name of the mapping weights Weights of the mapping matrix Entries are normalized to 1 number of entries must match the number of reference beads in a coarse grained bead name Name of molecule in coarse grained representation topology Section containing definition of coarse grained topology of molecule cg beads Section defining coarse grained beads of molecule cg bead Definition of a coarse grained bead beads The beads section lists all atoms of the reference system that are mapped to this particular coarse grained bead The syntax is RESID RESNAME ATOMNAME the beads are separated by spaces mapping Mapping scheme to be used for this bead specified in section map
93. uld be used to fit a functional form to the potential Nevertheless special attention should be paid to units angles bondlengths The resulting curve can then be specified in the MD package used for simulation However most potentials don t allow an easy processing of this kind and tabulated potentials have to be used 4 1 Generating a topology file for coarse grained run WARNING This section describes experimental features The exact names and options of the program might change in the near future The section is specific to GROMACS support though a generalization for other MD packages is planned The mapping definition is close to a topology needed for a coarse grained run To avoid redundant work csg_gmxtopol can be used to automatically generate a gromacs topology based on an atomistic reference system and a mapping file At the current state csg gmxtopol can only generate the topology for the first molecule in the system If more molecule types are present a special tpr file has to be prepared The program can be executed by csg gmxtopol top topol tpr cg map xml out cgtop which will create a file cgtop top This file includes the topology for the first molecule including definitions for atoms bonds angles and dihedrals It can directly be used as a topology in GROMACS however the force field definitions atom types bond types etc still have to be added manually 4 2 Post processing of the potential The voTCA
94. ustrate how to overload the script which calls the sampling package The csg inverse script runs mdrun from the GROMACS package only on one cpu Our task will be to change the script so that GROMACS uses 8 cpus which is basically the same as adding mpirun options in inverse gromacs mdrun command First we find out which script calls mdrun csg call list grep gromacs The output should look as follows init gromacs initalize gromacs sh prepare gromacs prepare gromacs sh run gromacs run gromacs sh pressure gromacs calc pressure gromacs sh rdf gromacs calc rdf gromacs sh imc stat gromacs imc stat generic sh convert potential gromacs potential to gromacs sh the third line indicates the script we need If the output of csg call is not clear one can try to find the right script in sec 10 4 Alternatively check the folder csg installation share scripts inverse 37 38 CHAPTER 9 ADVANCED TOPICS for all available scripts Analyzing the output of CSg call cat run gromacs we can conclude that this is indeed the script we need as the content in shorted form is critical mdrun Now we can create our own SCRIPTDIR add a new script there make it executable and overload the call of the script mkdir p SCRIPTDIR csg call quiet show run gromacs SCRIPTDIR my run gromacs sh chmod 755 SCRIPTDIR my run gromacs sh echo run gromacs my run gromacs sh SCRIPTDIR csg table Please note that my run gromacs sh is
95. x min name step Bibliography 1 V R hle C Junghans A Lukyanov K Kremer and D Andrienko Versatile object oriented toolkit for coarse graining applications J Chem Theor Comp page accepted 2009 1 2 14 211 29 2 Berk Hess Carsten Kutzner David van der Spoel and Erik Lindahl Gromacs 4 Algorithms for highly efficient load balanced and scalable molecular simulation J Chem Theo Comp 4 3 435 447 2008 1 2 3 W G Noid J Chu G S Ayton V Krishna S Izvekov G Voth A Das and H C Andersen The multiscale coarse graining method 1 a rigorous bridge between atomistic and coarse grained models J Chem Phys 128 244114 JUN 2008 3 4 7 4 Dominik Fritz Vagelis A Harmandaris Kurt Kremer and Nico F A van der Vegt Coarse grained polymer melts based on isolated atomistic chains Simulation of polystyrene of dif ferent tacticities Macromolecules 42 19 7579 7588 2009 W Isch p Kremer J Batoulis T Burger Hahn Simulation of polymer melts i coarse graining procedure for polycarbonates Acta Polymerica 49 61 74 1998 4 5 19 6 D Reith M P tz and F Miiller Plathe Deriving effective mesoscale potentials from atomistic simulations J Comp Chem 24 13 1624 1636 2003 6 31 7 Ap Lyubartsev and Laaksonen Calculation of effective interaction potentials from radial distribution functions a reverse monte carlo approach Phys Rev E 52 4
96. xp BU q 2 11 where Z f exp 8U q dq is a partition function 8 1 kgT Once P q is known one can obtain the coarse grained potential which in this case is a potential of mean force by inverting the probability distribution P q of a variable q which is either bond length bond angle or torsion angle U q kgT ln P q 2 12 The normalization factor Z is not important since it would only enter the coarse grained potential U q as an irrelevant additive constant Note that the histograms for the bonds H r angles Ho 0 and torsion angles H p have to be rescaled in order to obtain the volume normalized distribution functions P r P9 and P o respectively _ He 0 sin P r Pa 0 Arr Poly Hey 2 13 where r is the bond length r 0 is the bond angle and q is the torsion angle bonded coarse grained potential can then be written as a sum of distribution functions U r 6 p U r Ue 0 Uolo 2 14 U q ksT In P 9 g 7 9 On the technical side the implementation of the Boltzmann inversion method requires smooth ing of U q to provide a continuous force Splines can be used for this purpose Poorly and un sampled regions that is regions with high U q shall be extrapolated Since the contribution of these regions to the canonical density of states is small the exact shape of the extrapolation is less important Another crucial issue is

VOTCA manual

Contents

Download Pdf Manuals

Related Search

Related Contents