Machine Learning-Guided Protein Engineering

Abstract

Recent progress in engineering highly promising biocatalysts has increasingly involved machine learning methods. These methods leverage existing experimental and simulation data to aid in the discovery and annotation of promising enzymes, as well as in suggesting beneficial mutations for improving known targets. The field of machine learning for protein engineering is gathering steam, driven by recent success stories and notable progress in other areas. It already encompasses ambitious tasks such as understanding and predicting protein structure and function, catalytic efficiency, enantioselectivity, protein dynamics, stability, solubility, aggregation, and more. Nonetheless, the field is still evolving, with many challenges to overcome and questions to address. In this Perspective, we provide an overview of ongoing trends in this domain, highlight recent case studies, and examine the current limitations of machine learning-based methods. We emphasize the crucial importance of thorough experimental validation of emerging models before their use for rational protein design. We present our opinions on the fundamental problems and outline the potential directions for future research.

Simplification of Corticosteroids Biosynthetic Pathway by Engineering P450BM3

“…The principle of enzyme engineering involves the generation of enzymes with new functionalities through rational design, serving as efficient catalytic components for the synthesis of target compounds . The CYP450 superfamily of heme-thiolate monooxygenase enzymes are typically anchored to microbial membrane structures and function physiologically in conjunction with the oxidoreductase enzyme CPR .…”

“…Acevedo-Rocha et al further discovered that interactions within the loops, helices, and β-strands in the substrate access channel of P450BM3 affect the efficiency of substrate entry into the heme center, thereby affecting the efficiency of hydroxylating steroids. Thus, engineering the substrate access channel of P450BM3 can enhance catalytic efficiency and diversify the oxidation of steroids. , Moreover, computational simulation techniques can be utilized to decipher the potential functions of individual residues within enzyme–substrate complexes, providing impetus for precision-driven enzyme engineering, , and computational simulation-driven refinement in P450 engineering can offer efficient targeted oxidation components for the two-step transformation of PG into hydrocortisone.…”

Perspectives on Computational Enzyme Modeling: From Mechanisms to Design and Drug Development

“…In recent years, the application of machine learning techniques has revolutionized the field of protein structure prediction, , protein design, ,,,, and enzyme engineering. ,− These techniques focus primarily on either structure prediction or activity prediction. On the structure side, achievements such as AlphaFold2 and RoseTTAFold2 have significantly improved the accuracy of protein structure prediction.…”

Opportunities and Challenges for Machine Learning-Assisted Enzyme Engineering

“…Recently, machine learning (ML) has emerged as a useful tool for enzyme engineering, both for the discovery of functional enzymes, which is the focus of the first section of this Outlook, and for navigating protein fitness landscapes for fitness optimization, which is the focus of the second section. We encourage readers to read other reviews summarizing recent advancements in these areas. − ML is particularly well suited for the challenges of enzyme engineering, as generative models can take advantage of patterns in known protein sequences and supervised models can learn from labels of protein properties such as various measures of fitness. In this Outlook, we explain existing methods where ML is used to assist enzyme engineering, and we propose ML-related research efforts that can have the most beneficial impact for engineering outcomes.…”