Abstract
<title>Abstract</title> <p>Antimicrobial resistance (AMR) among ESKAPE pathogens poses an escalating threat to global health as conventional antibiotic pipelines fail to keep pace with emerging resistance. Antimicrobial peptides (AMPs) have re-emerged as a promising alternative due to their broad-spectrum, membrane-targeted mechanism and low resistance-inducing propensity; however, traditional AMP discovery remains slow, costly, and difficult to scale. Here, we present a systematic in silico pipeline for generation, filtering, and structural prioritization of pathogen-targeted AMPs using BioAMPify, a conditional generative model trained on pathogen-specific AMP sequence distributions. From an initial library of 872 sequences generated against six ESKAPE pathogens, we applied sequential filtering comprising physicochemical constraints (length, charge, hydrophobicity, hydrophobic moment), redundancy reduction, Boman index-based protein-binding filtering, ML-based toxicity prediction, and structure-based analysis using AlphaFold3, DSSP assignment, and principal component analysis (PCA). This narrowed the pool to 31 non-toxic peptides with plausible membrane-active structures and, ultimately, 10 high-priority candidates with high helical content (≥ 80%), favorable amphipathic organization, and low protein-binding propensity. Feature-loading analysis indicated that candidate separation was primarily driven by hydrophobicity and protein-binding propensity, with amphipathic character contributing to structural stabilization, consistent with established membrane-disruption models for helical AMPs. These results demonstrate that a pathogen-conditioned generative model combined with multi-stage computational filtering can efficiently enrich for structurally plausible, membrane-active AMP candidates from a large sequence space. The resulting 10 high-priority peptides constitute a resource-efficient shortlist for future molecular dynamics validation and experimental testing, offering a scalable framework for early-stage AMP discovery against multidrug-resistant ESKAPE pathogens.</p>