Physics-based Modeling in the New Era of Enzyme Engineering

Enzyme engineering techniques optimize enzymes to synthesize value-added chemicals, degrade environmental pollutants, and improve therapeutics. The field is entering a new era characterized by the increasing integration of computational strategies. While bioinformatics and artificial intelligence (AI) have been extensively applied to accelerate the screening of function-enhancing mutants, physics-based modeling methods, such as molecular mechanics and quantum mechanics, serve as essential complements in engineering objectives where setting up high-throughput screening is difficult or where a deep understanding of unknown physical principles is crucial. In this perspective, we discuss the enormous, untapped potential of physics-based modeling in guiding the next step of computational enzyme engineering. We first explore the paradigm of physics-based design principles wherein insights from natural, efficient enzymes are applied to recommend beneficial mutations in silico. We examine current development of high-throughput molecular modeling workflows that aid enzyme engineering campaigns through large-scale virtual applications of design principles. We then emphasize how physics-based modeling empowers AI techniques through enriching data expressiveness and interpretability. Finally, we proposed unmet challenges for the next step advancement of computational tools for enzyme engineering.