Register

 

In this webinar, we show a number of techniques to automate the GROMACS tools and run them in parallel in high-performance computing (HPC) environments.

GROMACS is a popular molecular dynamics (MD) software package. Its MD engine is well suited for HPC environments. However, the ancillary GROMACS tools are more difficult to use on clusters: they often require interactive input, they generally do not work in parallel, and they must be scripted from the command line shell (Bash) rather than using Python or other programming languages.

These limitations can be overcome. With Bash "here documents", we can feed input to the GROMACS tools to avoid interactive prompts. With the GNU parallel program or SLURM job arrays, we can run the GROMACS tools in parallel on one or more compute nodes. For complex problems, we can use the MDAnalysis Python package to read and analyse GROMACS structures and trajectories, and to write index groups that can then be used with the GROMACS tools. We also discuss how trajectory pre-processing can significantly speed up analysis.