Universal design of peptide binders for therapeutic and diagnostic modalities
Abstract
Structure prediction technology has transformed protein design, yet key challenges remain, particularly in engineering functional interactions. Many proteins exert their effects through binding interactions, but designing these effectively remains a major hurdle. While most efforts focus on large, stable proteins, shorter peptides offer advantages such as lower production costs and reduced steric hindrance, making them attractive therapeutic candidates.
Previously, we developed EvoBind, a design pipeline that leverages Monte Carlo search with AlphaFold2 to generate peptide binders (https://www.biorxiv.org/content/10.1101/2022.07.23.501214v1). EvoBind is now used worldwide, with its loss function independently validated through laboratory affinity analysis (https://www.nature.com/articles/s42256-023-00691-9). We now seek to enhance EvoBind by designing entire peptide sequences through gradient descent optimization within a trained structure prediction network (https://www.biorxiv.org/content/10.1101/2023.07.04.547638v1).
We will apply EvoBind to design peptide binders targeting key receptors in cancer, metabolic diseases, and other therapeutic areas, evaluating their efficacy through surface plasmon resonance. The goal is to create a diverse peptide library for precision medicine, enabling new treatments for oncology, diabetes, obesity, and beyond. By targeting disease-specific receptors, EvoBind-designed peptides could serve as both diagnostics and therapeutics, dynamically adapting to evolving molecular profiles.
Beyond binding affinity, EvoBind now enables the design of agonists and antagonists, broadening its scope for hormone modulation, immune regulation, and targeted protein degradation. By expanding and experimentally validating these capabilities, we aim to drive the next generation of AI-powered peptide drugs, revolutionizing treatment across multiple disease areas.
