HPC Run

ChemFlow was designed to profit from High Performance Computer (HPC) resources throught SLURM or PBS schedullers (SGE comming soon).

Usage

HPC resources may be requested through –slurm, –pbs flags, followed by an appropriate TEMPLATE indicated by the –header flag.

Sample TEMPLATES

Here are examples for this header file. One must always provide the HEADER for SLURM and PBS and edit them carefully.

PBS - Using the public queue, asking 2 nodes, 28 cores/node, for 2h

#! /bin/bash
#PBS -q public
#PBS -l nodes=2:ppn=28
#PBS -l walltime=2:00:00
#PBS -N myjob
...

SLURM - Using the gpu queue, asking 1 node with 4 GPUs, for 10 minutes.

#! /bin/bash
#SBATCH -p gpu
#SBATCH -n 1
#SBATCH -t 00:10:00
#SBATCH --gres=gpu:4
#SBATCH --job-name=myjob
...

Additional configuration may needed such as loading the compiler, MPI and CUDA libraries. Also any specifics for proprietary software, such as Amber or Gaussian that may differ from one’s workstation installation.

# Load modules
module load compilers/gnu
module load mpi/openmpi-3.0
module load compilers/cuda-10.1

# Path to amber.sh replace with your own
source ~/software/amber18/amber.sh

# Load Gaussian
module load gaussian
...

Tip

Seek assistance from the system administrator to optimally configure the TEMPLATE files.

Sample command lines:

LigFlow:

Prepare compounds with RESP charges, using 28 cores/node and through the SLURM scheduller.

LigFlow \
-p myproject  \
-l compounds.mol2 \
--resp \
-nc 28 \
--slurm --header TEMPLATE.slurm

DockFlow:

Dock compounds using AutoDock Vina. Using 16 cores/node through the PBS scheduller.

# AutoDock Vina, 16 cores/node, PBS
DockFlow \
-p myproject \
--protocol vina \
-r receptor.mol2 \
-l compounds.mol2 \
--center 31.50 13.74 24.36 \
--size 11.83 14.96 12.71 \
-sf vina \
-nc 16 \
--pbs --header TEMPLATE.pbs

ScoreFlow:

Standard Minimization and Molecular Dynamics Simulaton in explicit solvent, with RESP charges for the ligand, followed by MM/PBSA binding energy. Using 4 cores and 4 GPUs/node, and double precision CUDA.

ScoreFlow \
-p myproject \
--protocol MMPBSA \
-r receptor.pdb \
-l compounds.mol2 \
-sf mmpbsa \
--resp --md --water \
--cuda-double \
-nc 4 \
--slurm --header TEMPLATE.slurm

Tip

Use the –write-only flag to run a test before launching High Throughput calculations.

Warning

Be aware that HPC systems comonlly limit the amount of submitted jobs, choose your options wisely.