Standalone charge assignment from Espaloma framework. https://doi.org/10.1039/D2SC02739A
We recomend installing espaloma_charge via conda-forge:
$ mamba create -n espaloma -c conda-forge espaloma_chargeIf you plan on using openff-toolkit, then install that as well when creating your espaloma_charge environment for optimal dependency solving:
$ mamba create -n espaloma -c conda-forge espaloma_charge openff-toolkitWe also have espaloma_charge on pypi:
$ pip install torch torch-geometric espaloma_chargeOption0: Assign charges to rdkit molecule.
>>> from rdkit import Chem; from espaloma_charge import charge
>>> molecule = Chem.MolFromSmiles("N#N")
>>> charge(molecule)
array([0., 0.], dtype=float32)Assign charges to your favorite molecule in
Option1: Use with openff-toolkit(installation required)
>>> from openff.toolkit.topology import Molecule
>>> from espaloma_charge.openff_wrapper import EspalomaChargeToolkitWrapper
>>> toolkit_registry = EspalomaChargeToolkitWrapper()
>>> molecule = Molecule.from_smiles("N#N")
>>> molecule.assign_partial_charges('espaloma-am1bcc', toolkit_registry=toolkit_registry)
>>> molecule.partial_charges
<Quantity([0. 0.], 'elementary_charge')>Option2: Use as Command Line Interface to write antechamber-compatible charges.
$ espaloma_charge -i in.mol2 -o in.crg
$ antechamber -fi mol2 -fo mol2 -i in.mol2 -o out.mol2 -c rc -cf in.crg For a reproducible development environment, we use pixi. It mixes conda-forge packages (rdkit, openff-toolkit, pytorch) with a PyPI torch-geometric, and installs espaloma_charge itself in editable mode.
$ pixi install -e dev
$ pixi run -e dev pytest -v espaloma_chargeA minimal dataset for training the model can be found in https://doi.org/10.5281/zenodo.17308526
Coming soon.
If you are using this little tool in your pipeline, please consider citing:
@Article{D2SC02739A,
author ="Wang, Yuanqing and Fass, Josh and Kaminow, Benjamin and Herr, John E. and Rufa, Dominic and Zhang, Ivy and Pulido, Iván and Henry, Mike and Bruce Macdonald, Hannah E. and Takaba, Kenichiro and Chodera, John D.",
title ="End-to-end differentiable construction of molecular mechanics force fields",
journal ="Chem. Sci.",
year ="2022",
volume ="13",
issue ="41",
pages ="12016-12033",
publisher ="The Royal Society of Chemistry",
doi ="10.1039/D2SC02739A",
url ="http://dx.doi.org/10.1039/D2SC02739A"}
@misc{https://doi.org/10.48550/arxiv.2302.06758,
doi = {10.48550/ARXIV.2302.06758},
url = {https://arxiv.org/abs/2302.06758},
author = {Wang, Yuanqing and Pulido, Iván and Takaba, Kenichiro and Kaminow, Benjamin and Scheen, Jenke and Wang, Lily and Chodera, John D.},
keywords = {Machine Learning (cs.LG), Chemical Physics (physics.chem-ph), FOS: Computer and information sciences, FOS: Computer and information sciences, FOS: Physical sciences, FOS: Physical sciences},
title = {EspalomaCharge: Machine learning-enabled ultra-fast partial charge assignment},
publisher = {arXiv},
year = {2023},
copyright = {Creative Commons Attribution 4.0 International}
}