Loading [MathJax]/jax/output/HTML-CSS/config.js
Volume 7, Issue 2
Benchmarking Universal Machine Learning Force Fields with Hydrogen-Bonding Cooperativity

Xinping Feng, You Xu & Jing Huang

Commun. Comput. Chem., 7 (2025), pp. 152-160.

Published online: 2025-06

[An open-access article; the PDF is free to any online user.]

Export citation
  • Abstract

Machine learning force fields (MLFFs) offer a promising balance between quantum mechanical (QM) accuracy and molecular mechanics efficiency. While MLFFs have shown strong performance in modeling short-range interactions and reproducing potential energy surfaces, their ability to capture long-range cooperative effects remains underexplored. In this study, we assess the ability of three MLFF models — ANI, MACE-OFF, and Orb — to reproduce cooperative interactions arising from environmental induction and dispersion, which are essential for many biomolecular processes. Using a recently proposed framework, we quantify hydrogen bond (H-bond) cooperativity in N-methylacetamide polymers. Our results show that all MLFFs capture cooperativity to some extent, with MACE-OFF yielding the closest agreement with QM data. These findings highlight the importance of evaluating many-body effects in MLFFs and suggest that H-bond cooperativity can serve as a useful benchmark for improving their physical fidelity.

  • AMS Subject Headings

  • Copyright

COPYRIGHT: © Global Science Press

  • Email address
  • BibTex
  • RIS
  • TXT
@Article{CiCC-7-152, author = {Feng , XinpingXu , You and Huang , Jing}, title = {Benchmarking Universal Machine Learning Force Fields with Hydrogen-Bonding Cooperativity}, journal = {Communications in Computational Chemistry}, year = {2025}, volume = {7}, number = {2}, pages = {152--160}, abstract = {

Machine learning force fields (MLFFs) offer a promising balance between quantum mechanical (QM) accuracy and molecular mechanics efficiency. While MLFFs have shown strong performance in modeling short-range interactions and reproducing potential energy surfaces, their ability to capture long-range cooperative effects remains underexplored. In this study, we assess the ability of three MLFF models — ANI, MACE-OFF, and Orb — to reproduce cooperative interactions arising from environmental induction and dispersion, which are essential for many biomolecular processes. Using a recently proposed framework, we quantify hydrogen bond (H-bond) cooperativity in N-methylacetamide polymers. Our results show that all MLFFs capture cooperativity to some extent, with MACE-OFF yielding the closest agreement with QM data. These findings highlight the importance of evaluating many-body effects in MLFFs and suggest that H-bond cooperativity can serve as a useful benchmark for improving their physical fidelity.

}, issn = {2617-8575}, doi = {https://doi.org/10.4208/cicc.2025.90.02}, url = {http://global-sci.org/intro/article_detail/cicc/24185.html} }
TY - JOUR T1 - Benchmarking Universal Machine Learning Force Fields with Hydrogen-Bonding Cooperativity AU - Feng , Xinping AU - Xu , You AU - Huang , Jing JO - Communications in Computational Chemistry VL - 2 SP - 152 EP - 160 PY - 2025 DA - 2025/06 SN - 7 DO - http://doi.org/10.4208/cicc.2025.90.02 UR - https://global-sci.org/intro/article_detail/cicc/24185.html KW - Machine learning force field, cooperative effects, self-assembly, neural network potential, hydrogen bond. AB -

Machine learning force fields (MLFFs) offer a promising balance between quantum mechanical (QM) accuracy and molecular mechanics efficiency. While MLFFs have shown strong performance in modeling short-range interactions and reproducing potential energy surfaces, their ability to capture long-range cooperative effects remains underexplored. In this study, we assess the ability of three MLFF models — ANI, MACE-OFF, and Orb — to reproduce cooperative interactions arising from environmental induction and dispersion, which are essential for many biomolecular processes. Using a recently proposed framework, we quantify hydrogen bond (H-bond) cooperativity in N-methylacetamide polymers. Our results show that all MLFFs capture cooperativity to some extent, with MACE-OFF yielding the closest agreement with QM data. These findings highlight the importance of evaluating many-body effects in MLFFs and suggest that H-bond cooperativity can serve as a useful benchmark for improving their physical fidelity.

Feng , XinpingXu , You and Huang , Jing. (2025). Benchmarking Universal Machine Learning Force Fields with Hydrogen-Bonding Cooperativity. Communications in Computational Chemistry. 7 (2). 152-160. doi:10.4208/cicc.2025.90.02
Copy to clipboard
The citation has been copied to your clipboard