molecular-simulations 0.3.0

A small package for building molecular systems using the AMBER force field and deploying OpenMM simulations on HPC clusters using Parsl.

molecular-dynamics

Utilities for MD simulation building, running and analysis.

Capable of building molecular dynamics systems using the AMBER forcefield ecosystem. Available forcefields include: fixed-charge forcefields (ff19SB for proteins, OL21 for DNA, and OL3 for RNA), ff15ipq polarizable forcefield, and the ability to parameterize small molecules using GAFF2. In all explicit systems OPC water is used with the exception of polarizable systems where a polarizable water model is used (). Untested support for implicit solvent systems as well.

Installation

Easily installed via pip. PyPI page: https://pypi.org/project/molecular-simulations/

pip install molecular-simulations

Package details

Simulations are performed using OpenMM version 7.7 due to CUDA versioning on ALCF Polaris. In theory supports OpenMM >= 8.2 but this remains to be tested. All MD simulations are deployed on a PBS scheduler via Parsl but with some minor tweaking of the Parsl configuration object they can be deployed on theoretically any HPC cluster/job scheduler. Jobs can also be run locally without the need for Parsl (example coming soon).

Simulation analysis can also be performed using the provided Analyzer object (molecular_simulation.analysis.Analyzer). This leverages the MDAnalysis API for performing analysis which is both easy to work with and easily parallelized across CPU resources.

examples/ contains sample execution scripts

examples/run_builder.py

Leverages classes in build/build_amber.py for building implicit solvent, explicit solvent and biomolecule + small molecule ligand systems. Can handle multicomponent systems out of the box so long as the correct force fields are loaded. Additionally supports the polarizable protein ff amber15ipq although this remains untested.

examples/run_sim_analysis.py

Leverages classes found in analysis/analyzer.py for performing analysis using the MDAnalysis library. This was chosen due to its ongoing development and ease of object-oriented framework as well as straightforward parallelization.

examples/run_omm.py

Sets up OpenMM simulation objects and performs simple equilibrium simulation of both implicit and explicit solvent simulations using Parsl. Configured by default to run on the Argonne ALCF Polaris supercomputer but can be adapted for any scheduler on any cluster by modifying the Parsl configuration.

See Parsl documentation for further examples of how to run on other HPC resources. https://parsl.readthedocs.io/en/stable/