In silico Enzyme Engineering: Molecular Simulation and Machine Learning for Rapid Screening and Discovery

Enzymes are among the most promising tools available to address ongoing and emerging needs for sustainable, efficient, and cheap catalysis, and will play a central role in the forthcoming bioeconomy. However, enzymes are highly intricate macromolecules whose catalytic properties are often difficult to predict and engineer. Atomistic simulations of enzymes are an important tool for revealing the complex mechanisms underlying enzyme function, but they remain underutilized in screening for improved engineered variants due to a host of technical hurdles. In this talk I will describe contributions to the use of atomistic simulations for the discovery and characterization of unknown enzyme reaction mechanisms, and I will show how this knowledge can be extended through the use of creative simulations to greatly simplify the task of computationally screening the effects of enzyme mutations. I will then demonstrate how machine learning approaches to enzyme engineering can benefit from simulation data, both to directly speed up the search for promising enzyme variants and to complement other sources of information about enzymes, such as experiments and proteomics. Taken as a whole, this work lays the foundation for a new subdiscipline of computational enzyme engineering that synergizes with existing and emerging approaches to this important problem.