AIPAMPDS: an AI platform for antimicrobial peptide design and screening

Abstract

Antimicrobial resistance (AMR) presents a pressing global health crisis, particularly in the post-pandemic era, underscoring the urgent need for innovative therapeutics. Antimicrobial peptides (AMPs) hold promises for combating multidrug-resistant infections but are hindered by challenges in clinical translation, including limited hemocompatibility and inefficient design processes. To overcome these obstacles, we developed AIPAMPDS, an AI-driven computational platform that integrates a deep generative model with multi-stage screening to design highly active, non-haemolytic AMPs. The platform features two key modules: (1) a Generative Pre-trained Transformer (GPT)-based framework for target-specific AMP generation, enabling rapid exploration of novel peptide sequences, and (2) a multi-neural network screening pipeline for dual evaluation of antimicrobial activity and toxicity. By systematically evaluating both AI-generated and natural candidate AMPs, we identified ten lead peptides for experimental validation, of which 90% exhibited negligible haemolytic activity while maintaining robust antimicrobial potency. AIPAMPDS bridges computational design and therapeutic application, significantly advancing AMP discovery and providing a powerful tool for next-generation antimicrobial development. The platform is freely accessible via a web server at https://amps.pianlab.cn/AIPAMPDS/. GRAPHICAL ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=94 SRC="FIGDIR/small/644774v1_ufig1.gif" ALT="Figure 1"> View larger version (35K): org.highwire.dtl.DTLVardef@14e9a20org.highwire.dtl.DTLVardef@168e811org.highwire.dtl.DTLVardef@5bd9beorg.highwire.dtl.DTLVardef@344c34_HPS_FORMAT_FIGEXP M_FIG C_FIG