Large language models enable high-throughput mining and generation of antimicrobial peptides against clinical superbugs

The discovery of novel antimicrobial peptides (AMPs) against clinical superbugs is urgently needed to address the ongoing antibiotic resistance crisis. Here, a pre-trained protein large language model (LLM), ProteoGPT was established and further developed into multiple specialized sub-LLMs to assemble a sequential pipeline. This pipeline enables rapid screening across hundreds of millions of peptide sequences, ensuring potent antimicrobial activity and minimizing cytotoxic risks. Through transfer learning, we endowed the LLMs with different domain-specific knowledge to simultaneously achieve high-throughput mining and generation of AMPs within a unified methodological framework. Notably, both mined and generated AMPs exhibit reduced susceptibility to resistance development in ICU-derived superbugs, and comparable or superior therapeutic efficacy and anti-inflammatory properties compared to clinical antibiotics, without causing organ damage and disrupting gut microbiota.