Transferable Implicit Solvent Machine Learning Potential for Drugs and Proteins Approaching Ab Initio Accuracy

2026-07-12

AI Summary

This work addresses the limitations of existing machine learning interatomic potentials, which suffer from slow inference speeds that hinder microsecond-scale biomolecular simulations, and traditional implicit solvent models, whose accuracy is constrained by reliance on empirical force fields. The authors propose TWIN, a fully equivariant graph neural network–based, transferable implicit aqueous solvent machine learning potential trained end-to-end exclusively on ab initio quantum chemical calculations and experimental data (from crystallography and NMR), without any empirical force field components. TWIN achieves near-DFT accuracy while offering two orders of magnitude faster inference than explicit-solvent DFT-level machine learning potentials. It significantly outperforms current implicit solvent and coarse-grained models across multiple ab initio and experimental benchmarks, demonstrating broad applicability to drug-like molecules, peptides, and proteins.

Abstract

Machine learning interatomic potentials (MLPs) have revolutionized atomistic modeling, offering the potential to replace traditional methods like Density Functional Theory (DFT). However, inference time of MLPs is orders of magnitude slower than that of classical force fields, hindering real-world applications for biomolecular systems that require timescales of microseconds and beyond. Implicit solvent MLPs can address this issue, but are faced with data challenges associated with coarse-grained modeling. Consequently, previous approaches relied on empirical force field data, thereby inherently limiting the MLP's accuracy. Here, we introduce the Transferable Water Implicit Network (TWIN), an implicit water MLP parametrized entirely by an Equivariant Graph Neural Network and trained solely on ab initio and experimental labels. We demonstrate TWIN's transferability across drug-like molecules, peptides, and proteins, achieving excellent results on ab initio and experimental crystallographic and NMR benchmarks, consistently outperforming previous machine-learning-based implicit solvent or coarse-grained models. Furthermore, TWIN closely matches DFT-based explicit solvent MLPs while providing a two-order-of-magnitude faster timestep evaluation, paving the way for efficient ab initio-level modeling of biomolecular systems in aqueous environments.

Problem

• machine learning interatomic potentials
• implicit solvent
• ab initio accuracy
• biomolecular simulation
• computational efficiency

Innovation

• implicit solvent
• machine learning potential
• equivariant graph neural network
• ab initio accuracy
• transferability