Enhanced-Sampling-in-AMBER-tutorial

Tutorials on AMBER based Metadynamics and Gaussian Accelerated MD (GaMD) simulations on Chignolin. For running AMBER metadynamics, please refer to the first half of this video

General information

1. chignolin.pdb, chignolin.prmtop and chignolin.rst are the input files required for the AMBER simulation.

2. md.in is the AMBER run file, which contains the details of the simulation. The parameters that you can vary are
   a. ntpr = 500 ! write energy information every 500 steps to log file
ntwx = 500 ! write coordinates every 500 steps to a trajectory file
ntwr = 500 ! write restart file every 500 steps to corresponding file

   b. nstlim = 2000000 ! perform 2000000 MD steps
dt = 0.002 ! use time steps of 2fs

Metadynamics

1. Apart from the above input files, a plumed.dat file is also important, as it has the information about the paramters to be used in a metadynamics simulation. An example file has been given in this repo.
2. To run PLUMED on AMBER, the following commands are required in md.in:
   plumed = 1 ! Plumed will run
plumedfile="plumed.dat" ! Name of the plumed.dat which AMBER has to access
3. To run the simulation after preparing all the files, run sbatch run.sh. The run.sh should contain the AMBER run command, the export PLUMED_KERNEL command and all the relevant modules and SBATCH headers. An example file has been provided.
4. 2 PLUMED output files will be produced. COLVAR file will have the information about the collective variables and the bias added, while HILLS file will have the details of the gaussian bias potentials that were added over the simulation.
5. Run plumed sum_hills --hills HILLS --mintozero --min 0 --max 10 --bin 200 to get the fes.dat file that has the reweighted free energy profile.

Gaussian Accelerated Molecular Dynamics (GaMD)

1. To run GaMD on AMBER, additional commands have to be added to md.in (Please see the !! GAMD settings section of md.in in the GaMD folder of this repository)
2. To run the simulation after preparing all the files, run sbatch run.sh. The run.sh should contain the AMBER run command and the relevant modules and SBATCH headers. An example file has been provided.
3. 2 GaMD output files will be produced. gamd.log is the file that stores all the potential energy information through the simulation, and gamd-restart.dat are the GaMD parameters that the simulation uses.
4. Run the command cpptraj -i analysis.cpptraj to get the rmsd.dat and/or rg.dat (Radius of gyration) information from the simulation. You have to do conda activate AmberTools23 in order to access the cpptraj command.
5. For reweighting to get the free energy surface (FES):
   a. Get weights.dat from gamd.log using the command awk 'NR%1==0' gamd.log | awk '{print ($8+$7)/(0.001987*300)" " $2 " " ($8+$7)}' > weights.dat
   b. Get output.dat from rmsd.dat and rg.dat (If using 2 progress coordinates for a 2D surface) using the command awk 'NR==FNR{a[NR]=$2; next} {print a[FNR], $2}' rmsd.dat rg.dat > output.dat
   c. Run the command ./reweight-2d.sh 50 50 0.1 0.1 output.dat 300, where 50 50 are the cutoffs for both progress coordinates, 0.1 0.1 are the bin spacings, and 300 is 300 K.

Exercise

After running the Metadynamics and/or GaMD simulation, write Python codes to read the free energy data files, and plot the free energy profiles (Hint: NumPy and Matplotlib should be enough ; use np.genfromtxt to read the files).
Which Enhanced Sampling method performs better?