Search papers, labs, and topics across Lattice.
This study evaluates the performance of an electrostatic embedding scheme for relative binding free energy (RBFE) calculations in protein-ligand systems, addressing limitations of fixed-charge approximations in classical force fields. By training a TensorNet2 model on a large dataset, the authors demonstrate that electrostatic embedding significantly improves accuracy metrics for the TYK2 target, achieving a root mean square error (RMSE) of 0.45 kcal/mol, while yielding mixed results for other targets. The findings highlight the importance of target-specific evaluations in assessing the effectiveness of machine learning approaches in computational chemistry.
Electrostatic embedding can dramatically enhance binding free energy predictions for specific targets, but its effectiveness varies significantly across different systems.
Alchemical relative binding free energy (RBFE) calculations are limited by the fixed-charge approximation of classical force fields. Hybrid machine learning interatomic potential/molecular mechanics (MLIP/MM) schemes correct ligand strain, but under mechanical embedding still describe ligand--environment electrostatics with static point charges. Electrostatic embedding schemes coupling machine-learned charges to the MM environment have been proposed and validated against QM/MM for simple systems, but not tested in a production alchemical workflow. We take the electrostatic embedding scheme of Semelak et al.\ and evaluate it on protein--ligand RBFE. We trained a TensorNet2 model, \texttt{AceFF-2-RESP-1}, on $10^{6}$ conformations from the AceFF dataset, jointly predicting energies, forces and Restrained Electrostatic Potential (RESP) charges. We chose RESP over MBIS for commensurability with the AMBER-family force field it couples to. The predicted charges enter the short-range direct-space part of the particle mesh Ewald sum, with Thole damping to prevent polarization catastrophes during alchemical transformations. We tested the scheme across five targets from the Wang et al.\ benchmark set, fixed in advance by a prior study, with three replicates per edge and matched protocols. Electrostatic embedding improved every accuracy and correlation metric for TYK2 ($\Delta\Delta G$ RMSE $0.86 \rightarrow 0.45$~kcal/mol against GAFF2), but performed comparably to the classical and mechanical-embedding baselines for CDK2, thrombin, p38 and JNK1. Standard single-molecule energy and charge benchmarks were not good predictors of this target-dependent outcome. TYK2 combined good $\Delta\Delta G$ accuracy with the lowest force error on the Schr\"odinger benchmark, but this pattern did not hold for the other targets.
