Answers are grounded in this paper and cite their sources.
Antimicrobial peptide screening requires both activity classification and pathogen-specific potency estimation. Here we present GeoMIC, a two-stage framework combining protein language model embeddings with predicted peptide geometry. GeoMIC achieved an AUROC of 0.844 on XUAMP and minimum inhibitory concentration (MIC) prediction correlations of 0.794 for Staphylococcus aureus and 0.797 for Escherichia coli. Capacity-matched ablations supported complementary contributions from predicted structure. Retrospective ranking recovered more known active peptides than the evaluated comparator, supporting further evaluation of GeoMIC for experimental candidate prioritization.
No takes yet. Share an insight, caveat, or question.
Tang et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: