Artificial intelligence for predicting mutation effects on enzyme kinetics

Predicting how mutations affect enzyme kinetics, especially catalytic turnover (k_cat), remains a major challenge in enzymology. Reliable predictions could help identify promising variants early and reduce experimental screening in enzyme-engineering campaigns. Advances in artificial intelligence (AI) are expanding the tools available to address this challenge. Current AI approaches include molecular simulations using machine learning potentials, supervised models trained on experimental kinetics data, and generative sequence models that rank mutation effects. This review summarizes recent progress, remaining limitations, and opportunities for improving kinetic prediction, with an emphasis on studies that evaluate multiple variants and systematically assess the relationship between computational predictions and experimental measurements.