Skip to content

Latest commit

 

History

History
185 lines (135 loc) · 6.26 KB

README.md

File metadata and controls

185 lines (135 loc) · 6.26 KB
Raw

amber_protocols

Repository of computational biochemistry protocols for the Amber package

Set up AMBERPROTOCOLS environment variable for ease of use

export AMBERPROTOCOLS=<path-to-this-repo>

Protocols consist of various scripts and configuration files for running and analyzing molecular dynamics with Amber. Protocols are organized by macro-area within the protocol directory:

  • protein: protocols for simple protein-solvent simulations.
  • membrane_protein: protocols for protein simulations with lipid membranes.
  • solvate: protocols to solvate system binding sites using RISM.

Resources

Example workflow

  1. Copy the desired protocol folder to your working directory.
cp -r $AMBERPROTOCOLS/protocols/protein/protocol .

  1. (optional) If needed, edit the protocol files

  2. Edit the loader script load.sh with the appropriate atom selection masks for your system. Atom selections are needed to specify which atoms should be restrained during the relaxation steps. Note, the & character must be escaped, e.g.: & → \&.

  3. Run load.sh to load the protocol and the run.sh template in your working directory. This also creates a local copy of the loop_step and run_steps runner scripts.

./protocol/load.sh

  1. Edit the run.sh script with the appropriate parameters for your system:

  2. Run the simulation

sbatch run.sh

Restart/Extend simulations

  • To restart a simulation, simply re-run the sbatch command; the checkpoint (.ok) files will cause already completed steps to be skipped.
  • To extend a simulation, simply increase the value of NSTEP variable in the run.sh script (RUN_PROD section) and restart the simulation.

WARNING: be careful of the 0-padding. A 0-padded suffix specifying the step number will be appended automatically to the filenames generated by the MD. By default, the length of the padding is the same of the digits of NSTEPS. For example, NSTEPS=10 will generate suffixes from 01 to 10, but NSTEPS=100 will generate suffixes from 001 to 100. This might be a problem when extending a simulation requires increasing the number of digits in NSTEPS (e.g., 50 ⇒ 100). This will have the effect that previously generated checkpoint and restart files won't be recognized, causing the simulation to restart from the beginning. The easiest solution to this is to add a suffix to NAME (e.g., my_md ⇒ my_md_ext) or to run the simulation in a separate folder. With this approach, INPCRD variable should now be modified to point to the last step of the previous simulation (you can prepend absolute/relative path if needed) and NSTEPS should now correspond to the number of steps to be added to the previous steps.

If you plan on extending your simulation, it is advised to 0-pad NSTEPS in advance, e.g., using NSTEPS=050 allows restarting up to step 999.

Run replicas

  1. Edit the run.sh script with the appropriate parameters for your system: prepend a reference to the $REPLICA_DIR environment variable to indicate input/output paths that should be relative to each replica (i.e., output base NAME and INPCRD input coordinates). The $REPLICA_DIR variable is set automatically by run_replica.sh (see the next step below).
NAME=$REPLICA_DIR/MD_equil
INPCRD=$REPLICA_DIR/last_step.rst7
  1. Copy the run_replica.sh script into your working directory
cp $AMBERPROTOCOLS/utils/run_replica.sh .
  1. Edit the local copy of run_replica.sh with the desired parameters:
NREPLICAS=6           # Number of replicas
BASENAME=replica      # Basename of replica directory
  1. Run the simulations in parallel. The script will handle the creation of directories for each replica if they don't exist and will spawn a job for each replica by launching the run.sh script with the appropriate values of $REPLICA_DIR.
./run_replica.sh

Restart individual replicas

# re-launch job for replica_3
env REPLICA_DIR=replica_3 sbatch run.sh

To extend replicas, modify run.sh as explained in Restart/Extend simulations. To add replicas, simply add to NREPLICAS within the run_replica.sh script file Completed jobs will be launched but will terminate immediately as long as checkpoint (.ok) files exit.

Utils

Restarting and Stopping

Checkpoint files will be written for each completed step with the name <basename>_step_<stepname or stepnum>.ok, if such file exists, the corresponding step will be skipped, so that restarting a simulation only requires to re-run run.sh.

At any point, if a file with the name <basename>_step_<stepname or stepnum>.stop is found, the execution will be halted before that step.

loop_step

loop_step: prmtop inpcrd step nrep basename

Repeat a step (mdin) nrep times, with the output of each run being used as the input for the next one. step can be a list of two filenames, if so, the first input file will be used only for the first step. Useful to create "starter" input files.

run_steps

run_steps: prmtop inpcrd steps basename

Launch a series of steps (mdin) in sequence, with the output of each step being used as the input for the next one. steps should be a list of filenames, separated by newline, spaces, or commas. e.g., $( find . -name "*.in" | sort )

rep_step

rep_step: step rep

Copy a configuration (mdin) file rep times, adding a progressive suffix. Useful to repeat the same step multiple times using run_steps.

Analysis

Some cpptraj recipes for common tasks.