With the rise of drug-resistant superbugs, antimicrobial peptides (AMPs) are promising next-generation therapeutics. We developed an AI-driven peptide discovery platform integrating CNN prediction, modified wGAN generation, and toxicity screening to design novel low-toxicity peptides. Several candidates showed antimicrobial, antifungal, and anticancer activity with minimal hemolysis. This framework accelerates peptide drug discovery while reducing development time and cost.

We developed an AI-integrated therapeutic peptide platform combining large-scale databases, deep learning prediction, generative AI, and toxicity evaluation. Using PC6 encoding with CNN models, our system achieved up to 95% accuracy across AMP, ACP, AVP, and AFP tasks, outperforming traditional ML methods. We also introduced the first hemolysis prediction model integrating peptide sequence and concentration, enabling rapid design of low-toxicity, high-activity peptides.

This technology can be applied to antimicrobial drugs, smart wound dressings, antibacterial coatings, animal health, and precision medicine. The AI-integrated platform enables rapid design of low-toxicity functional peptides, reducing drug development cost and time while improving translational efficiency. It also has strong commercialization potential in biomedical, pharmaceutical, aquaculture, livestock, and biomaterial industries.