Artificial Intelligence-assisted Mining of Polyethylene Terephthalate Hydrolases

Polyethylene terephthalate (PET) hydrolases efficiently hydrolyze the ester bonds in PET, converting it into valuable monomers or oligomers, offering a sustainable biological solution to global PET plastic pollution. However, the large-scale development of high-performance PET hydrolases remains challenging due to limitations in traditional enzyme resource mining methods, including low throughput and lengthy cycles. Recent advances in artificial intelligence (AI) provide novel methodologies to overcome these challenges. This review systematically summarizes how AI empowers the high-throughput screening of PET hydrolases from mass biological databases, while allowing the accurate prediction of enzyme structures and functions. Furthermore, it critically analyzes AI-driven strategies for enzyme molecular engineering and highlights the emerging frontier of AI-assisted de novo enzyme design. By systematically evaluating the advantages and challenges of AI models in the research of PET hydrolases, this review provides an integrated technical framework and theoretical foundation to guide future innovation in enzyme mining and plastic biodegradation. • AI enables PET hydrolase discovery, prediction, engineering and de novo synthesis. • The AI models of PET hydrolase mining were comprehensively introduced • Practical guidance on model selection, established and emerging methods are mentioned. • Several challenges in the application of AI remain to be addressed. • Directions on enhancing performance of AI models are created.