Advances in Zero-Shot Prediction-Guided Enzyme Engineering Using Machine Learning

Abstract

The advent of machine learning (ML) has significantly advanced enzyme engineering, particularly through zero-shot (ZS) predictors that forecast the effects of amino acid mutations on enzyme properties without requiring additional labeled data for the target enzyme. This review comprehensively summarizes ZS predictors developed over the past decade, categorizing them into predictors for enzyme kinetic parameters, stability, solubility/aggregation, and fitness. It details the algorithms used, encompassing traditional ML approaches and deep learning models, emphasizing their predictive performance. Practical applications of ZS predictors in engineering specific enzymes are discussed. Despite notable advancements, challenges persist, including limited training data for ZS predictors and the necessity to incorporate environmental factors (e.g., pH, temperature) and enzyme dynamics into these models. Future directions are proposed to advance ZS prediction-guided enzyme engineering, thereby enhancing the practical utility of these predictors.