Intelligent De Novo Design of Novel Antimicrobial Peptides against Antibiotic-Resistant Bacteria Strains

Because of the growing number of clinical antibiotic resistance cases, novel antimicrobial peptides (AMPs) are attractive candidates for next-generation antibiotics. This study trained a Wasserstein generative adversarial network with gradient penalty (WGAN-GP) on known AMPs to generate novel AMP candidates. After in silico evaluation, eight candidates (GAN-pep 1–8) were selected by an AMP AI classifier and synthesized for validation. Disc diffusion testing and minimum inhibitory concentration (MIC) determinations assessed antibacterial effects. Seven of the eight synthesized GAN-designed peptides displayed antibacterial activity. GAN-pep 3 and GAN-pep 8 showed broad-spectrum effects and activity against antibiotic-resistant strains, including methicillin-resistant Staphylococcus aureus (MRSA) and carbapenem-resistant Pseudomonas aeruginosa (CRPA). GAN-pep 3 had low MICs across all tested bacteria. This demonstrates an efficient approach to discover AMPs effective against general and antibiotic-resistant bacteria and suggests that similar strategies can be used to design other functional peptides for rapid wet-lab validation.