Artificial Intelligence and Machine Learning in Protease Engineering and Optimisation

Protein engineering and optimisation using machine learning (ML) and artificial intelligence (AI) have entirely transformed the objectives of protein engineering, enabling the rapid discovery of proteins and the improvement of enzyme properties. Computational methods have been paired with conventional methods such as directed evolution, and ML is one of the core components of the plan for finding more suited starting variants, calculating the effect of mutations, and optimising fitness landscapes with the help of high-throughput data analysis. The supervised models have been effective in predicting the impact of mutations and annotating protein function. Generative adversarial networks (GANs) and variational autoencoders (VAEs) have been utilised to develop novel enzymes with diverse catalytic activities in solution. In this regard, AI-based processes have been instrumental in promoting the stability and catalysis of proteases and avoiding all major obstacles, such as substrate selectivity and unselective proteolytic cleavage. The latest approaches to achieving improved characterisation are substrate profiling and cleavage activity assays, which optimise engineered proteases for specific uses in state-of-the-art protein sequencing methods, resulting in high-fidelity biomolecular analysis. A combination of AI and ML with protease engineering is expected to provide an opportunity to discover new design principles for enzymes that were previously unimaginable, leading to significant changes in industrial biocatalysis, therapeutics, and biomolecular engineering. This review describes the application of AI/ML in protease optimisation and prospects, influencing protease engineering.