Alchemical protein-ligand free energy benchmark study using Perses and espaloma-0.3.0

This repository includes scripts to validate espaloma-0.3.0 force field with relative alchemical protein-ligand binding free energy infrastrcuture, Perses. This repository is part of espaloma-0.3.0-manuscript.

Description

Here, we compare the relative alchemical protein-ligand binding free energy calculation accuracy using espaloma-0.3.0, espaloma-0.2.2, and openff-2.1.0 force fields against a custom protein-ligand benchmark dataset which was original taken from the OpenFF protein-ligand-benchmark.

Proteins are parameterized with Amber ff14SB and solvated with TIP3P water model with Joung and Cheatham monovalent counterions to neuralize the system. Small molecules are parameterized with either espaloma-0.3.0, espaloma-0.2.2, or openff-2.1.0 force field. Additional experiments are conducted where both the proteins and small molecules are parameterized with espaloma-0.3.0.

Manifest

• experiment/: Stores directories and scripts to run Perses
  • cdk2/
    • espaloma-0.2.2/
    • espaloma-0.3.0rc6/
    • espaloma-0.3.0rc6-complex/
    • openff-2.1.0/
  • mcl1/
    • espaloma-0.3.0rc6/
    • espaloma-0.3.0rc6-complex/
    • openff-2.1.0/
  • p38/
    • espaloma-0.3.0rc6/
    • espaloma-0.3.0rc6-complex/
    • openff-2.1.0/
  • tyk2/
    • espaloma-0.2.2/
    • espaloma-0.3.0rc6/
    • espaloma-0.3.0rc6-complex/
    • openff-2.1.0/
  • script/: Scripts to run the benchmark using Perses and analyze the results
    • run_benchmark.py
    • benchmark_analysis.py
• figures/: Stores scripts to plot figures
  • 01-plotall/: Plot free energy calculation results for all targets
  • 02-compare-plot/: Compare the first and second Perses runs
• envs/: Stores conda environment files
  • environment-0.2.4.yaml: Conda environment to run Perses with espaloma-0.2.2 to parameterize small molecules
  • environment-0.3.0.yaml: Conda environment to run Perses with openff-2.1.0 and espaloma-0.3.0 to parameterize small molecules
  • environment-0.3.0-v3.yaml: Conda environment to run Perses with espaloma-0.3.0 that parameterize both small molecules and proteins

Note

• espaloma-0.3.0rc6 refers to espaloma-0.3.0
• espaloma-0.2.2 is the first generation espaloma model described in the original paper of espaloma

Environment

Core dependencies are perses 0.10.1 and modified version of openmmforcefield 0.11.0 (commit hash: 6d2c3dcd33d9800a32032d28b6b2dca92f348a43) to support espaloma-0.3.0. A modified version of perses 0.10.1 (commit hash: 0d069fc1cf31b8cce1ae7a1482c3fa46bc1382d2) is required to run Perses to parameterize both small molecules and proteins with espaloma-0.3.0. All figures are plotted using a modified version of cinnabar 0.3.0 (commit hash: de7bc6623fb25d75848aa1c9f538b77cd02a4b01) to support arbitrary tick frequency when plotting the alchemical free energy calculation results.

Citation

If you find this helpful please cite the following:

@misc{takaba2023machinelearned, title={Machine-learned molecular mechanics force field for the simulation of protein-ligand systems and beyond}, author={Kenichiro Takaba, Iván Pulido, Pavan Kumar Behara, Chapin E. Cavender, Anika J. Friedman, Michael M. Henry, Hugo MacDermott Opeskin, Christopher R. Iacovella, Arnav M. Nagle, Alexander Matthew Payne, Michael R. Shirts, David L. Mobley, John D. Chodera, Yuanqing Wang}, year={2023}, eprint={2307.07085}, archivePrefix={arXiv}, primaryClass={physics.chem-ph} }