Accurate quantum-chemical analysis of protein-protein interactions (PPIs) requires a statistically sound treatment of conformational fluctuations in addition to electronic structure accuracy. Although molecular dynamics (MD) simulations have been combined with the fragment molecular orbital (FMO) method in previous studies, most approaches rely on a limited number of time-sampled snapshots and simple averaging, without explicitly accounting for underlying conformational populations. Here, we present an MD + FMO framework that incorporates structural clustering and population-weighted ensemble averaging to overcome this limitation. Large MD ensembles are compressed into a small number of representative conformational clusters, and FMO interaction analyses are performed for each representative structure. Inter-fragment interaction energies (IFIEs) are then evaluated as population-weighted ensemble averages, enabling efficient and physically meaningful incorporation of finite-temperature conformational statistics. The method is applied to the HER2-Pertuzumab antigen-antibody complex, including wild-type and point mutants with experimentally characterized binding affinities. The population-weighted FMO results reproduce experimental binding trends and, through residue-level interaction analysis, reveal how conformational redistribution and local interaction networks cooperatively modulate binding energetics, highlighting the roles of loop flexibility and salt-bridge stability. This approach provides a general framework for ensemble-aware quantum-chemical investigation of PPIs.
Enomoto et al. (2026) studied this question.