Generating parameters using RESP charges from GAMESS-US

From CUC3
Jump to navigation Jump to search

GAMESS

GAMESS-US is currently installed in ~csw34/gamess on Mek-quake, Clust and Volkhan and this version should be used for all GAMESS jobs. It is possible to write GAMESS jobs by hand, but simpler to use the Ante_RED.pl script which can be found in ~khs26/scripts/RED/. This script has been altered from the default RED scripts to include deriation of electrostatic potentials and to stop optimisation by GAMESS (which is time-consuming and less effective than in-house tools).

Syntax for using Ante_RED.pl is:

perl ~khs26/scripts/RED/Ante_RED.pl [input file with extension]

This should write Gaussian (.com) and GAMESS (.inp) input files, as well as a p2n and pdb file (check the output PDB file to make sure it looks right, often Ante_RED can interpret input PDB files incorrectly if you have extra columns that shouldn't be there). You need to manually edit the charge (ICHARG) and spin multiplicity (MULT) in the .inp file according to your particular molecule before running GAMESS.

Now check the following (GAMESS is quite fussy):

  • You have a ~/scr directory
  • This ~/scr directory does not contain a .dat file with the same name as your system (e.g. if your input PDB was SIA.pdb, there is no SIA.dat file)
  • You have specified the correct charge
  • The output PDB is correct and consistent with the GAMESS .inp file
  • You have allocated sufficient memory to GAMESS (this can be changed in the .inp file, the MWORDS keyword specifies the allocated memory in megawords)

Once all these checks have been completed, then you can run GAMESS. The default syntax requires you to specify a version, but Chris has hardcoded in the relevant version number and so the syntax is as follows:

~csw34/gamess/rungms [.inp input file, without extension] [number of processors] > [GAMESS output file (typically .log extension)]

N.B. The number of processors available to the queue and the number of processors specified must be consistent.

The error/output files generated by the job are generally not terribly useful in explaining any errors in the job, so look at the bottom of the GAMESS output file to see whether the job terminated properly, and if not, what the problem was.

From here there are two possible options:

  • Use the RED script to generate RESP charges
  • Use the GAMESS to RESP script to generate RESP charges

GAMESS to RESP script

A modified version of this script (originally available here) is at ~khs26/scripts/gamess_to_resp_scripts/gmstoresp.sh. The modified version takes syntax in the form:

gmstoresp.sh [absolute pathname of the original input pdb with extension] < [GAMESS output file with extension] > [Antechamber prepin file]

This script generates a .prepin file using Antechamber, so it could take a few minutes for large molecules. There is no error-checking for divcon, so make sure to check that Antechamber worked properly. From there you can run parmchk and generate topology files and coordinate files as normal with Amber. One easy way to check for problems is to look at the .prepin file in a text editor and check that symmetry equivalent atoms (e.g. methyl hydrogens) are symmetry equivalent (and there are no ridiculous charges) as Antechamber should ensure this.

RED script

I'll fill this in later... --khs26 13:01, 21 August 2008 (BST)

Back to AMBER