Difference between revisions of "ensmin.pl"
Revision as of 14:29, 12 August 2006
usage: ensmin.pl [options] intag outtag options: [-par CHARMMparams] [-opt file[:file]] [-l refPDB min:max[=min:max ...]] [-[no]conslim] [-limforce value] [-limsel=ca|cb|cab|heavy] [-cons [ca|cb|cab|heavy] self|refpdb min:max[_force][=...]] [-run [from:]to] [-dir workdir] [-natpdb pdbFile] [-[no]compress] [-update frq] [PARALLELoptions] [-log file]
This script is used to generate structure ensemble structures
through minimization of PDB files. Many options are similar
to <docmark>minCHARMM.pl</docmark>, the script for single
minimization runs, and <docmark>enslatsim.pl</docmark>, the
script for generating ensembles from lattice simulations.
Two tags are required as argument to ensmin.pl: The first one identifies the ensemble structures that are used for input at each run and have to be present already, the second one is used for the output after the minimization run is complete.
The ensemble directory can be given with -dir if it is not the current directory. The number of desired runs is specified using -run either followed by a single number for the total number of runs or by an explicit interval.
A reference PDB file for comparison with the native conformation may be given through -natpdb but is not required. For loop or fragment modeling the range of residues that are being modeled and a template PDB file with the structure outside the modeled residues are specified using -l. Like the native PDB structure this data may already be present in the ensemble configuration file and does not need to be given in that case.
Although this script will work in a serial fashion, one would probably want to take advantage of parallel architectures for a larger number of simulations. The available parallel options are listed <A HREF="http://mmtsb.scripps.edu/cgi-bin/parallelopt">here.</A>
The remaining options are used to control minimization runs and are kept in a file tag.options for later reference and subsequent executions of this script on the same ensemble set. They are very similar to the options in <docmark>minCHARMM.pl</docmark> with the exception of restraints for loop modeling:
In <docmark>minCHARMM.pl</docmark> the option -l switches on gradual restraints outside a given fragment list based on distance in sequence space. An alternative is to use -cons with a restraint list generated by <docmark>vicinity.pl</docmark>. Such a list will restrain residues surrounding the loop region in two shells with different force constants to allow some readjustment of residues in direct contact with the loop. In ensmin.pl a fragment list, given with -l, is not just used for restraints but also to limit RMSD and other calculations to the loop region. In typical loop modeling applications this option would always be specified even if the restraints are given through -cons. A second option -conslim is therefore necessary to request explicitly that the fragment list is also used to generate restraints. However, if a fragment list is given and no other restraint list is given with -cons this flag is turned on automatically, unless -noconslim is given. The options -limsel and -limforce to specify the atom selection and maximum force constant used for automatically restraints from the fragment list. If any of these options is given the conslim is also turned on automatically.
- usage information
ensmin.pl -dir data -run 5 -par steps=200,tol=1E-2,param=19 lat vacmin
performs 5 minimization runs starting from the lat ensemble structures and stores them under the tag vacmin Each run is done in vacuum (default) over 200 steps with the adopted-basis Newton-Raphson scheme or until an energy tolerance of 1E-2 is reached. CHARMM19 parameters are used.
ensmin.pl -dir data -run 6:20 -cpus 3 lat vacmin
adds more minimization runs for indices 6 through 20 to an existing ensemble in the data directory. The simulations are run in parallel on 3 CPUs.
ensmin.pl -dir data -cpus 2 -conslim -limforce 1.0 -limsel ca -noslowrest lat vacmin
minimizes a set of ensemble structures that have been generated previously through lattice simulations in a loop modeling application (see