Training a force field for proteins and small molecules from scratch

Abstract

Force fields for molecular dynamics are usually developed manually, limiting their transferability and making systematic exploration of functional forms challenging. We developed a graph neural network that assigns all force field parameters for diverse molecules using continuous atom typing. The freely-available model, called Garnet, was trained on quantum mechanical, condensed phase and protein nuclear magnetic resonance data without the use of existing parameters. The resulting force field shows comparable performance to current force fields on small molecules, folded proteins, protein complexes and disordered proteins. It shows similar results to popular approaches for relative binding free energy predictions across a range of targets. Assessing different functional forms shows that the double exponential potential is a flexible and accurate alternative to the Lennard-Jones potential. Garnet provides a platform for automated, reproducible force field discovery that brings the benefits of machine learning to classical force fields.

A transcriptional reporting system for sensing interactions of small molecules with proteins
Koder Dagher, Minqi Pan, Basilius Sauter +4
ChemRxiv29 Sep 2026PreprintLicence not statedGeneral chemistryChemRxiv
From Proteins and Molecules to Cats and Dogs: Visualization as Scalable Minimum Spanning Trees
Alejandro Flores Sepúlveda, Maarten Boneschansker, Daniel Probst +1
ChemRxiv4 Sep 2026PreprintCC BYGeneral chemistryChemRxiv
Full-data accuracy with fewer labels for training and fine-tuning machine-learning force fields
Sheng Bi, Yi-Ze Wang, Jun Cheng
arXiv26 Oct 2026PreprintarXiv licenceFull textPhysical chemistry Condensed matterarXiv
Understanding multi-fidelity training of machine-learned force-fields
John L. A. Gardner, Hannes Schulz, Jean Hélie +2
Machine Learning Science and Technology21 Jul 2026Peer-reviewedCC BYFull text8 citedMaterials & polymersOpenAlex
MPID Polarizable Force Field: GPU Implementation and Biomolecular Simulations
Qiaozhu Tan, Andrew C Simmonett, Bernard R. Brooks +1
ChemRxiv20 Sep 2026PreprintLicence not statedGeneral chemistryChemRxiv

Cite

BibTeX RIS

BibTeX

@article{blancogonzalez2026training,
  title = {{Training a force field for proteins and small molecules from scratch}},
  author = {Alexandre Blanco-González and Thea K. Schulze and Evianne Rovers and Joe G. Greener},
  journal = {Chemical Science},
  year = {2026},
  month = sep,
  doi = {10.1039/d6sc02874h},
  url = {https://doi.org/10.1039/d6sc02874h},
  note = {CC BY 3.0},
}

RIS

TY  - JOUR
TI  - Training a force field for proteins and small molecules from scratch
AU  - Alexandre Blanco-González
AU  - Thea K. Schulze
AU  - Evianne Rovers
AU  - Joe G. Greener
PY  - 2026
DA  - 2026/09/17
JO  - Chemical Science
DO  - 10.1039/d6sc02874h
UR  - https://doi.org/10.1039/d6sc02874h
AB  - Force fields for molecular dynamics are usually developed manually, limiting their transferability and making systematic exploration of functional forms challenging. We developed a graph neural network that assigns all force field parameters for diverse molecules using continuous atom typing. The freely-available model, called Garnet, was trained on quantum mechanical, condensed phase and protein nuclear magnetic resonance data without the use of existing parameters. The resulting force field shows comparable performance to current force fields on small molecules, folded proteins, protein complexes and disordered proteins. It shows similar results to popular approaches for relative binding free energy predictions across a range of targets. Assessing different functional forms shows that the double exponential potential is a flexible and accurate alternative to the Lennard-Jones potential. Garnet provides a platform for automated, reproducible force field discovery that brings the benefits of machine learning to classical force fields.
SN  - 2041-6520
SN  - 2041-6539
ER  -

Abstract

Related

Cite