Molecular simulations of food proteins: Assessing the basis and limits of interpretation

Molecular dynamics (MD) simulations are increasingly applied to food proteins, often beyond the systems and conditions where the underlying models are well characterized. This review examines force-field benchmark coverage in food-protein MD and discusses the inferential distance between MD-derived quantities and experimental measurements. It also discusses recurring challenges and approaches beyond conventional atomistic MD. Direct force-field benchmarking remains sparse across food-protein systems and conditions, leaving model-dependent uncertainty incompletely characterized. Beyond model uncertainty, interpretive scope depends on how directly MD-derived quantities relate to the claimed outcome. Cross-scale explanations of higher-level functionality are therefore better treated as hypothesis-generating interpretations rather than direct evidence of mechanisms. Among recurring challenges, limited assessment of sampling reliability, particularly the frequent absence of reported independent replication, appears the most immediately addressable. Overall, this review emphasizes the need to consider both the scope of claims and the adequacy of the simulations supporting them in food-protein MD.