Universal design of peptide binders for therapeutic and diagnostic modalities
Abstract
Structure prediction has revolutionised protein design, yet functional binder design, especially of small peptides, remains a major challenge. Short peptides offer advantages such as low production cost, membrane permeability, and reduced steric hindrance, making them ideal therapeutic candidates.
We previously developed EvoBind, a peptide binder design framework that leverages AlphaFold-based predictions in a Monte Carlo setting. EvoBind is now widely adopted, and its design capacity for producing novel linear and cyclic peptide high-affinity binders has been validated (https://www.nature.com/articles/s42004-025-01601-3). To expand the design space, we created RareFold, a new structure predictor supporting noncanonical amino acids (NCAAs), and extended EvoBind into EvoBindRare, enabling the generation of linear and cyclic binders with enhanced chemical diversity.
These tools now allow the design of not only binders but also functional agonists and antagonists, broadening their utility in hormone signalling, immune modulation, and targeted degradation. We aim to apply this technology to therapeutic targets in cancer, diabetes, and obesity, building a precision peptide library for next-generation diagnostics and treatments.
