Knowledge Distillation of a Protein Language Model Yields a Foundational Implicit Solvent Model
Journal of Chemical Theory and Computation · 28.07.2026 · 22 (14) · 7382-7393
DOI: 10.1021/acs.jctc.6c00574 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Chemical Theory and Computation, online erschienen am 03.07.2026. Verfasst von Justin Airas und 1 weiteren Autor:innen. Beteiligte Einrichtung: Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts02139, United States. Hauptthemen laut MEDLINE: Solvents, Molecular Dynamics Simulation, Intrinsically Disordered Proteins, Proteins. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 140 Literaturangaben. Der Volltext ist frei in PubMed Central verfügbar. Gefördert von: NIGMS NIH HHS, National Science Foundation Graduate Research Fellowship Program, National Institute of General Medical Sciences.
Abstract (Originalsprache)
Implicit solvent models (ISMs) promise to deliver the accuracy of explicit solvent simulations at a fraction of their computational cost. However, despite decades of development, their accuracy has remained insufficient for many critical applications, particularly for simulating protein folding and the behavior of intrinsically disordered proteins. Developing a transferable, data-driven ISM that overcomes the limitations of traditional analytical formulas remains a central challenge in computational chemistry. Here, we address this challenge by introducing a novel strategy that distills the evolutionary information learned by a protein language model, ESM3, into a computationally efficient graph neural network (GNN). We show that this GNN potential, trained on effective energies from ESM3, is robust enough to drive stable, long time-scale molecular dynamics simulations. When combined with a standard electrostatics term, our hybrid model accurately reproduces protein folding free energy landscapes and predicts the structural ensembles of intrinsically disordered proteins. This approach yields a single, unified model that can be transferred across both folded and disordered protein states, resolving a long-standing limitation of conventional ISMs. By successfully distilling evolutionary knowledge into a physical potential, our work delivers a foundational ISM poised to accelerate the development of predictive large-scale simulation tools.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Substanzen
Autor:innen und Institutionen
- Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts02139, United States.
ORCID-Profile: Justin Airas, Bin Zhang
Förderung
- NIGMS NIH HHS
- National Science Foundation Graduate Research Fellowship Program
- National Institute of General Medical Sciences
Zitieren
APA
Airas, J., & Zhang, B. (2026). Knowledge Distillation of a Protein Language Model Yields a Foundational Implicit Solvent Model. Journal of Chemical Theory and Computation, 22(14), 7382-7393. https://doi.org/10.1021/acs.jctc.6c00574
Vancouver
Airas J, Zhang B. Knowledge Distillation of a Protein Language Model Yields a Foundational Implicit Solvent Model. Journal of Chemical Theory and Computation. 2026;22(14):7382-7393. doi:10.1021/acs.jctc.6c00574
BibTeX
@article{Airas2026,
author = {Airas, Justin and Zhang, Bin},
title = {Knowledge Distillation of a Protein Language Model Yields a Foundational Implicit Solvent Model},
journal = {Journal of Chemical Theory and Computation},
year = {2026},
volume = {22},
number = {14},
pages = {7382--7393},
doi = {10.1021/acs.jctc.6c00574}
}
EUVIMED