Machine-learning force-field scoring rivals free-energy perturbation for congeneric ligand ranking across public benchmarks
- 1SandboxAQ
Abstract
Relative binding free-energy (RBFE) methods such as free-energy perturbation (FEP) set the accuracy standard for ranking congeneric ligands in structurebased drug design, but their computational cost limits throughput. Machine-learned force fields (MLFFs) now approach quantum-chemical accuracy at a small fraction of that cost, raising the question of whether they can recover much of RBFE’s ranking accuracy while remaining computationally efficient. We introduce a static-score methodology called taqo and benchmark 10 MLFF variants spanning 6 model families (UMA, MACE, Orb, eSEN, AIMNet2, AllScAIP), together with a classical molecular-mechanics force field, on a uniform staticpocket protocol, correlated against experiment across 22 benchmark systems.We find that the ten MLFFs cannot be resolved from one another at this sample size, while every one of them significantly outranks the classical force field reference, at a few seconds per ligand on a single GPU. Against published OpenFE and FEP+ results on the same systems and the same ligand sets, taqo is statistically indistinguishable from OpenFE and significantly below FEP+, at roughly three orders of magnitude lower cost. On the congeneric JACS subset taqo and FEP+ are statistically indistinguishable. On the same benchmark poses taqo outperforms classical docking scores and is statistically indistinguishable from deep-learning affinity models, without any affinity-specific training. Together these establish MLFF interaction-energy scoring as a practical high-throughput approach for lead optimization.
Related
- Accuracy and Efficiency Benchmarks of Pretrained Machine Learning Potentials for Molecular Simulations
- Beyond Organic Molecules: MetalLipoDB, a Lipophilicity Dataset for Metal Complexes and Machine Learning Benchmarks
- Finite-temperature dynamic stability separates foundation machine-learning interatomic potentials that harmonic benchmarks rank equally
- State-specific perturbation theory for configuration interaction singles: Theoretical analysis, benchmarks, and improvement with size-intensive level shifts
- A transferable perturbation proteomics framework for phenotype- 1 based annotation and antiproliferative potency scoring of antineoplastic agents
BibTeX
@misc{ryczko2026machinelearning,
title = {{Machine-learning force-field scoring rivals free-energy perturbation for congeneric ligand ranking across public benchmarks}},
author = {Kevin Ryczko and Sarah Maier and Amogh Sood and Patrick Rowe and Harish Ramadas and Lorenzo Boninsegna and Justin Overhulse and Hunter La Force and Mikayla Darrows and Shiji Zhao and Andrew Wildman and Mary Pitman and Romelia Salomon Ferrer and Andrea Bortolato},
howpublished = {ChemRxiv},
year = {2026},
month = sep,
doi = {10.26434/chemrxiv.15008810/v1},
url = {https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008810/v1},
note = {CC BY-NC},
}
RIS
TY - UNPB
TI - Machine-learning force-field scoring rivals free-energy perturbation for congeneric ligand ranking across public benchmarks
AU - Kevin Ryczko
AU - Sarah Maier
AU - Amogh Sood
AU - Patrick Rowe
AU - Harish Ramadas
AU - Lorenzo Boninsegna
AU - Justin Overhulse
AU - Hunter La Force
AU - Mikayla Darrows
AU - Shiji Zhao
AU - Andrew Wildman
AU - Mary Pitman
AU - Romelia Salomon Ferrer
AU - Andrea Bortolato
PY - 2026
DA - 2026/09/14
PB - ChemRxiv
DO - 10.26434/chemrxiv.15008810/v1
UR - https://chemrxiv.org/doi/full/10.26434/chemrxiv.15008810/v1
AB - Relative binding free-energy (RBFE) methods such as free-energy perturbation (FEP) set the accuracy standard for ranking congeneric ligands in structurebased drug design, but their computational cost limits throughput. Machine-learned force fields (MLFFs) now approach quantum-chemical accuracy at a small fraction of that cost, raising the question of whether they can recover much of RBFE’s ranking accuracy while remaining computationally efficient. We introduce a static-score methodology called taqo and benchmark 10 MLFF variants spanning 6 model families (UMA, MACE, Orb, eSEN, AIMNet2, AllScAIP), together with a classical molecular-mechanics force field, on a uniform staticpocket protocol, correlated against experiment across 22 benchmark systems.We find that the ten MLFFs cannot be resolved from one another at this sample size, while every one of them significantly outranks the classical force field reference, at a few seconds per ligand on a single GPU. Against published OpenFE and FEP+ results on the same systems and the same ligand sets, taqo is statistically indistinguishable from OpenFE and significantly below FEP+, at roughly three orders of magnitude lower cost. On the congeneric JACS subset taqo and FEP+ are statistically indistinguishable. On the same benchmark poses taqo outperforms classical docking scores and is statistically indistinguishable from deep-learning affinity models, without any affinity-specific training. Together these establish MLFF interaction-energy scoring as a practical high-throughput approach for lead optimization.
ER -