Agentic Discovery of Non-Canonical Antimicrobial Peptides with AMPGAN v3

AMPGAN v3, a multi-objective conditional GAN, has been released to enhance the design of antimicrobial peptides (AMPs) by incorporating D-amino acids and various chemical modifications. The model improves training stability through separate adversarial and activity-aware supervision across two discriminators, outperforming previous generative AMP models on external classifiers. This advancement is crucial for practitioners as it expands the generative capabilities for real-world peptide drug design, while the accompanying PepCraft framework facilitates end-to-end AMP discovery, aligning computational predictions with in vitro validation results.