Efficient Force Evaluation via Fragmentation: Toward Geometry Optimization of Protein-Ligand Systems with Quantum Mechanical Accuracy.

Abstract

Accurate evaluation of interatomic forces and interaction energies in large molecular systems, such as protein-ligand complexes or polymer-molecule assemblies, is essential for a wide range of applications in drug discovery and materials science. However, achieving a balance between computational efficiency and accuracy remains a major challenge: classical force fields offer high speed but limited precision, while quantum mechanical (QM) methods provide greater accuracy at the cost of poor scalability. In this work, we present a hybrid fragmentation-based approach, FragQMMM, that enables efficient and accurate force evaluation for large molecular systems. Specifically, crucial intermolecular interactions are treated at the semiempirical QM level (GFN2-xTB), while intramolecular forces are described using molecular mechanics (MM). This selective treatment greatly accelerates geometry optimization while preserving the essential interaction features. When applied to protein-ligand complexes, the method delivers a speed-up of roughly 10-100× compared to full QM calculations. The resulting geometries match the reference QM structures more closely than those obtained with pure MM, successfully reproducing the hydrogen-bond network and preserving the underlying structure-activity relationship.