Applications and limitations of AI tools in enzyme design
TL;DR
A comprehensive introduction and overview of several current artificial intelligence (AI)‐driven methods available for enzyme design, with a focus on reaction‐to‐sequence design, structure prediction, substrate scope prediction, engineering of stable variants, design of enzymes with non‐canonical amino acids, and de novo design is offered.
Abstract
Enzymes catalyze various different chemical reactions often with high efficiency and selectivity compared to synthetic catalysts. Advances in protein engineering over the past decades have allowed researchers to design enzymes, improving their catalytic performance and adapting them for specific or entirely novel chemical reactions. However, the need for the experimental validation of thousands of computational designs remains one of the major bottlenecks. Yet another restriction is our still limited knowledge about transition state architectures, effects of mutations, active‐site dynamics just to name a few. To overcome this, the combination of experimental and computational methods is essential, yet many experimentalists are facing significant obstacles when entering the field of computational enzyme design. To address these obstacles, this review offers a comprehensive introduction and overview of several current artificial intelligence (AI)‐driven methods available for enzyme design, with a focus on reaction‐to‐sequence design, structure prediction, substrate scope prediction, engineering of stable variants, design of enzymes with non‐canonical amino acids, and de novo design. Subsequently, this work serves as an accessible guide for experimental researchers with interest in learning how to use AI‐based computational methods in enzyme engineering.