StreaMD
Open-source toolkit for automated setup, execution, restart, and analysis of molecular dynamics simulations across many biomolecular complexes.
Scientific problem
Running molecular dynamics across thousands systems requires repetitive preparation, execution, restart, and analysis steps that are difficult to perform consistently by hand.
Motivation
StreaMD was developed to automate these technically repetitive stages without removing the scientific decisions needed to define an appropriate simulation protocol.
My role
- Principal developer of StreaMD; developed, tested, and maintained the automated MD workflow.
- Integrated system preparation, GROMACS execution, restart handling, trajectory processing, protein–ligand interaction analysis, and endpoint free-energy calculations.
- First author of the Journal of Cheminformatics publication describing the toolkit.
Methodology
- Automated preparation of proteins, ligands, cofactors, systems for GROMACS simulations.
- Minimization, equilibration, production MD, checkpoint-based restart, and trajectory extension.
- RMSD, RMSF, radius of gyration, trajectory fitting, and ProLIF interaction analysis.
- Automated MM/GBSA and MM/PBSA calculations and aggregation across complexes.
- CPU/GPU and distributed execution for multiple systems and replicas.
Results
- Validated on a benchmark of 624 protein–ligand complexes
- Used as part of the prospective CACHE Challenge #1 hit-finding workflow.
- Published in Journal of Cheminformatics (2024).
- 100+ GitHub stars.
Software
- Python
- GROMACS
- AmberTools
- MDAnalysis
- ProLIF
- gmx_MMPBSA
- Dask
- PBS
- SLURM