Nature Biomedical Engineering: HMD-AMP Protein Language Model Screens 37 Million+ Antimicrobial Peptide Candidates From Mammalian Microbiomes
A new Nature Biomedical Engineering paper introduces HMD-AMP, a protein language model-based approach that outperforms prior methods at identifying evolutionarily distant antimicrobial peptides. Applied to host and gut microbiome genomes of nine mammals, HMD-AMP revealed over 37 million predicted AMPs. Of 91 experimentally validated high-confidence sequences, 74 showed strong antibacterial activity and 48 were evolutionarily remote from known AMPs, including four with broad-spectrum activity at low toxicity.