Deep learning models are revolutionizing peptide drug discovery by enabling de novo design of peptide agonists with predicted biological activity. These AI-driven approaches are accelerating discovery timelines and expanding the range of targets that can be addressed with peptide therapeutics.
Recent advances in AI have shifted the focus towards structure prediction, generative design, and interaction modelling, significantly accelerating drug design and discovery[reference:159]. Deep learning architectures trained on peptide-receptor interaction data can propose novel peptide sequences with agonist activity, reducing the need for extensive screening.
Applications of deep learning in peptide design include generative models that propose novel sequences, predictive models that estimate binding affinity and activity, and optimization models that guide sequence modification. These models can be trained on experimental data, enabling them to learn the relationships between sequence and activity.
Recent developments include a generative artificial intelligence approach for peptide antibiotic optimization[reference:160], and a deep learning–attention framework for de novo generation of anti-diabetic peptide candidates[reference:161]. These approaches demonstrate the potential of AI to discover peptides with therapeutic potential.
Beyond sequence generation, AI is also being used to predict peptide properties including stability, solubility, and immunogenicity. These predictions guide the selection of candidates for synthesis and testing, reducing the number of peptides that need to be synthesized. At PeptideHub, we integrate AI-powered design with rapid synthesis and screening, enabling rapid identification of lead peptides for challenging targets.