Application of machine learning in the discovery of antimicrobial peptides: Exploring their potential for ulcerative colitis therapy

Abstract Ulcerative colitis (UC) is a chronic inflammatory bowel disease with rising global prevalence, yet existing treatments are not universally effective. Antimicrobial peptides (AMPs), produced by the immune system, have diverse antimicrobial and immune-regulatory functions, making them promising candidates for UC therapy. Using machine learning, we developed a machine learning-based prediction model to identify novel AMPs. The predicted peptides demonstrated significant biological activity in vitro and in vivo. In a dextran sulfate sodium-induced UC mouse model, engineered AMPs notably improved UC-related parameters, such as body weight, disease activity index (DAI), and colon length. These effects were likely mediated by modulation of Akkermansia muciniphila . This study highlights the potential of machine learning-identified AMPs as future therapeutic candidates for UC. Graphical abstract