• Welcome to the e-NMR Homepage >>


GROMACS use case example

This use case describes how to run GROMACS jobs on the eNMR grid.

A gzipped tar file containing all necessary files and an example of a job file for grid submission is provided for download. This use case provides a basis for executing any GROMACS script on the eNMR grid.

File: GridGMX-demo.tgz

Description of content:

This directory contains an example of running a short (protocollized) 1 ns MD simulation on he GRID, starting from a single pdb file.

Created by Tsjerk A. Wassenaar, Utrecht University, April 2009

Input files:

  • run-gmx.jdl: job file for grid submission.
    The job file contains the controls for running the script through the arguments session.

  • run.sh: the executable script that will be run on the GRID
    This script is a generic wrapper for automatically running complex programs on the grid that require many input files and scripts/programs (it will unpack input data, library files, protocols, scripts and other dependencies; run executable (from archive) with options given; pack output data and collect garbage and recycle bits).

  • The data (BRSV-G.pdb) and dependencies (MD.tgz) are sent to the GRID where they are unpacked if needed. Subsequently, a script from the dependency repository is executed, which may make calls to other scripts and programs. run.sh has a limited number of options that are parsed by the script itself. Other arguments are passed onto the script that is called. From the help (runs.sh -h):

  • File options:
    --run-data data.tgz :  Input data
    --run-lib lib.tgz :  Library files protocols and scripts
    --run-exec exec :  Executable
    --run-out out.tgz :  Output data and log files
    --run-log run.log :  Log file for capturing stdout/stderr
    --run-time :  Maximum run time, after which job will be terminated
    --clean :  Destroy input data after run

  • MD.tgz:
    This is the library containing the scripts required for setting up and running the simulation. This file does not contain Gromacs, which needs to be available on the CE. This file needs to be available regardless of the PDB file and generall will not require editing.

  • BRSV-G.pdb:
    This is a sample PDB file to run the demo. The same procedure should work for any PDB file adhering to the following requirements: There are no missing atoms or residues and there are no other molecules than protein and/or nucleic acids

 

Execution on the eNMR grid: (commands are in italics)

1) Login as an eNMR grid user:

voms-proxy-init -voms enmr.eu

2) Submit job file:

glite-wms-job-submit -o run-gmx.id -a run-gmx.jdl

3) Check status:

glite-wms-job-status -i run-gmx.id

4) When finished get back output:

glite-wms-job-output -i run-gmx.id --dir gmx-output

5) unpack results:

cd gmx-output
tar xvzf mdresults.tgz

Notes:

  • To run MD on another structure (e.g. protein.pdb):

    The options -o and --run-data,;5C which in the example are set to 'BRSV-G.pdb', should be changed if another pdb file is used for input. Use sed to edit the jdl file: sed -i 's/BRSV-G.pdb/protein.pdb/g' run-gmx.jdl