ParGaMD combines Gaussian Accelerated Molecular Dynamics (GaMD) with the Weighted Ensemble (WE) framework to enhance sampling efficiency in molecular simulations.

Simulation Steps:

1. Run a Conventional GaMD Simulation:

   • Navigate to the cMD directory.
   • Execute: sbatch run_cmd.sh
   • This generates gamd-restart.dat for ParGaMD setup.

2. Configure the WE Framework:

   • In the main directory, run: ./run_WE.sh
   • This copies gamd-restart.dat and bstate.rst to the appropriate WE directories.

3. Update west.cfg:

   • Set pcoord_len to nstlim/ntpr + 1 in west.cfg.
   • Ensure nstlim and ntpr in common_files/md.in are correctly specified.

4. Submit the ParGaMD Simulation:

   • Retrieve the cGaMD job ID: squeue -u <username>
   • Submit ParGaMD as a dependent job:

     sbatch --dependency=afterok:<jobid> run_WE.sh

   • Update NODELOC in env.sh to the current directory.

5. Post-Processing:

   • After completion, process data to obtain gamd.log and PC.dat:

     sbatch run_data.sh

   • Note: Submit run_data.sh on a compute node to avoid memory issues.

Reweighting for Free Energy Surfaces (FES):

1. Extract Weights:

   • Generate weights.dat from gamd.log:

     awk 'NR%1==0' gamd.log | awk '{print ($8+$7)/(0.001987*300)" "$2" "($8+$7)}' > weights.dat

2. Prepare Coordinates:

   • Combine PC1.dat and PC2.dat for a 2D surface:

     awk 'NR==FNR{a[NR]=$2; next} {print a[FNR], $2}' PC1.dat PC2.dat > output.dat

3. Generate Free Energy Surface:

   • Run the reweighting script:

     ./reweight-2d.sh 50 50 0.1 0.1 output.dat 300

     • 50 50: Cutoffs for both progress coordinates.
     • 0.1 0.1: Bin spacing for the coordinates.
     • 300: Temperature in Kelvin.

For detailed instructions and resources, refer to the ParGaMD GitHub repository.

By following these steps, you can effectively perform ParGaMD simulations to explore biomolecular dynamics with enhanced sampling and parallelization.

Sources:

1. For installing WESTPA see instructions at https://github.com/westpa/westpa
2. GamD information at Yinglong Miao, Victoria A. Feher, and J. Andrew McCammon Journal of Chemical Theory and Computation 2015 11 (8), 3584-3595 DOI: 10.1021/acs.jctc.5b00436 and https://www.med.unc.edu/pharm/miaolab/resources/gamd/
3. https://github.com/westpa
4. https://github.com/Sonti974948/ParGaMD (ParGaMD for protein folding example if you want to start with a tutorial and smaller system)