System Requirements

Hardware requirements

CS-FEP requires a standard computer with GPUs like NVIDIA GeForce RTX 2080 Ti or Tesla V100-SXM3-32GB.

Software requirements

OS Requirements

This package is supported for Linux. The package has been tested on the following systems:

• Linux: CentOs 7

Simulation Software

• Modified Amber with version >= 18. The source code file ti.F90 in AMBERHOME/src/pmemd/src was modified in order to avoid reordering the atoms of ligand.

• To modify Amber software, patch files are provided for amber18 and amber 20.
  • For amber18: use the ti.F90.patch and patch.sh in amber18_patch
  • For amber20: use the ti.F90.patch and patch.sh in amber20_patch

The cotent of patch.sh

#!/bin/bash
# cd the path ($AMBERHOME//src/pmemd/src) that the ti.F90 file exists.
patch -p0 < ti.F90.patch

The installation guide of Amber: http://ambermd.org/Installation.php

• Typical install time of Amber on a "normal" desktop computer: about 40 mins.

Analysis Code Dependencies

• The analysis code depends on the Python scientific stack.

numpy
pandas
pymbar
matplotlib

• Typical install time of python and required packages on a "normal" desktop computer: about 20 mins.

Running simulation

• Use this script to run the MD simulation run.sh

#!/bin/bash
ROOT=`pwd`
TARGET='cdk2_28-26_demo'
WINS='0.00 0.05 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 0.95 1.00'
SIDES='complex/cM2A complex/cM2B ligands/lM2A ligands/lM2B'
_CUDA_VISIBLE_DEVICES_=0
for side in $SIDES;do
    cd $ROOT/$TARGET/$side
    for win in $WINS;do
        cd $win
        sh submit.sh $_CUDA_VISIBLE_DEVICES_
    done
done

• Expected run time for demo's MD simulation on a "normal" desktop computer(with one gpu like Tesla V100-SXM3-32GB): 6-8 hours
• More input files can be found here: https://www.scidb.cn/en/s/NrUrAz

Analysis

• The demo's simulation out was provided in the cdk2_28-26.tar.bz2, use the following command to extract the data:

tar -zxvf cdk2_28-26.tar.bz2

• Use the in-house free energy calculation python script: analysis_free_energy.py
• aly.sh

#!/bin/bash
ROOT=`pwd`
FEP_PATH="$ROOT/cdk2_28-26/FEP"
mkdir -p analysis_out
cd analysis_out
python $ROOT/analysis_free_energy.py -d ${FEP_PATH} -f 0.25 -o fe_out.csv
cd ..

• Expected run time for demo's analysis on a "normal" desktop computer: 3-5 mins