A unified framework for potency-oriented AMP discovery via multi-modal learning and guided sequence synthesis

The rapid emergence of drug-resistant pathogens poses a critical threat to global health. With traditional antibiotics losing efficacy, antimicrobial peptides (AMPs) have gained attention for their unique mechanisms and lower resistance potential. We aimed to accelerate AMP discovery by proposing a closed-loop framework that combines AMP-Hunter (a shared-architecture discriminator for AMP classification and MIC prediction that integrates convolutional neural networks with graph neural networks), and AMP-Forge (a generator integrating multiple sequence alignment to select original candidates) and is guided by minimum inhibitory concentration (MIC)for latent space optimization and candidate selection. AMP-Hunter outperformed baseline models in both AMP classification and MIC prediction, achieving 95.82% accuracy and a 95.80% F1 score on the test set for classification, and an R2 of 0.9245 with an MAE of 0.2305 for MIC prediction. Guided by its predictions, AMP-Forge generated peptide sequences with lower MIC values and improved physicochemical properties associated with antimicrobial activity. Molecular dynamics simulations further provided in silico evidence supporting the antimicrobial potential of selected sequences by identifying stable membrane disruption and insertion behaviors consistent with membrane-targeting activity. Thus, the generation–screening–validation workflow enables reliable discovery of potent AMPs, and provides a practical strategy for rational peptide design, rapid prediction, and translational applications.Author summary: As the efficacy of many traditional antibiotics gradually diminishes, drug-resistant bacteria are posing an increasingly serious threat to global health. Antimicrobial peptides, which are short chains of amino acids capable of combating bacteria, offer a highly promising alternative; however, the process of discovering and designing these peptides is often time-consuming, expensive, and inefficient. To address this challenge, we have developed a novel closed-loop workflow comprising two core components that can be used both to screen and predict peptide activity and to generate more effective novel peptides. Our discriminator accurately identifies promising peptides and predicts their ability to inhibit bacterial growth, outperforming existing methods. Guided by these predictions, the new peptides designed by our generator not only exhibit superior antibacterial activity but also feature improved physical and chemical properties, suggesting improved safety- and stability-related properties. We further confirmed through molecular dynamics simulations that these new peptides effectively disrupt bacterial membranes. Our method accelerates the discovery of effective antimicrobial peptides and provides a practical pathway for peptide design, thereby helping to address the challenge of antimicrobial resistance.