From evolution to rational design: AI-driven engineering of safe and highly efficient food enzymes

As a representative area of industrial biocatalysis, food enzyme engineering plays a critical role in advancing the food industry. Recently, bioinformatics, molecular simulations, and artificial intelligence (AI) have converged into a powerful toolkit for the full spectrum of computational design, from the modification of existing enzymes to the de novo creation of novel biocatalysts. Moving beyond isolated discussions of individual tools, this review synthesizes key strategies across this landscape, from sequence-based methods that leverage evolutionary information to structure-based approaches for engineering food enzyme stability and catalytic function. A central focus is the transformative impact of AI, which not only guides molecular modification but also drives enzyme design through generative models. In addition, the safety and regulatory challenges posed by designed food enzymes are highlighted. Methods for evaluating allergenicity, potential toxicity, and the genetic safety of production strains are presented, and the future development trends of AI-driven safety assessment are discussed. By integrating computational design with experimental validation and safety evaluation, this review provides a cohesive framework to accelerate the development of high-performance, application-oriented, and safe food enzymes.