Antimicrobial peptides (AMPs) are promising candidates in the fight against antibiotic resistance. However, traditional screening methods are often complex and costly. Metagenome offers a powerful approach for discovering novel AMPs from environmental samples. In this study, the deep learning model AMP-CLIP was employed to screen for novel AMPs from metagenomes of livestock farm soil and feces, which are rich in microbial diversity. The model achieved a high prediction accuracy of 99%.To enhance model interpretability, SHAP analysis was performed, revealing that amino acids contributing most to antimicrobial activity predictions were predominantly positively charged residues, suggesting that... More >