[Artificial intelligence-based mining of antimicrobial peptides in the microbiome]

Antimicrobial peptides (AMPs) are small-molecule polypeptides with broad-spectrum antimicrobial activity that are induced by the innate immune system of organisms and constitute a crucial component of the innate immune defense. With the misuse of antibiotics and other antimicrobial agents, the problem of bacterial resistance has become increasingly severe. Naturally occurring AMPs derived from the microbiome, owing to their broad availability, stable physicochemical properties, relatively low propensity to induce resistance, and capacity to contribute to the maintenance of commensal microbiota homeostasis, are regarded as beneficial supplements and adjuvant therapeutic strategies to traditional antibiotics. In recent years, the rapid advancement of artificial intelligence (AI) technologies has facilitated their application in the field of drug discovery, thereby opening new avenues for the large-scale screening of AMPs and substantially accelerating the overall research progress. This review summarizes the developmental trajectory of AMP research methodologies from early approaches to the present day and provides a comparative overview of AI-based AMP screening tools and databases. In addition, the criteria for AMP screening and the recent progress in AI-assisted AMP development, both domestically and internationally, are systematically compiled. The aim of this review is to provide a systematic synthesis of the methodological framework for AI-based mining of microbiome-derived AMPs, thereby offering theoretical references and technical guidance for the efficient identification of novel AMPs. It is anticipated that this review will stimulate further consideration regarding AMP screening and design, and will promote advancements in the field of human health in the era of AI.