De novo design of high-affinity protein binders with AlphaProteo
Abstract
Computational design of protein-binding proteins is a fundamental capability with broad utility in biomedical research and biotechnology. Recent methods have made strides against some target proteins, but on-demand creation of high-affinity binders without multiple rounds of experimental testing remains an unsolved challenge. This technical report introduces AlphaProteo, a family of machine learning models for protein design, and details its performance on the de novo binder design problem. With AlphaProteo, we achieve 3- to 300-fold better binding affinities and higher experimental success rates than the best existing methods on seven target proteins. Our results suggest that AlphaProteo can generate binders "ready-to-use" for many research applications using only one round of medium-throughput screening and no further optimization.
Cite
@article{arxiv.2409.08022,
title = {De novo design of high-affinity protein binders with AlphaProteo},
author = {Vinicius Zambaldi and David La and Alexander E. Chu and Harshnira Patani and Amy E. Danson and Tristan O. C. Kwan and Thomas Frerix and Rosalia G. Schneider and David Saxton and Ashok Thillaisundaram and Zachary Wu and Isabel Moraes and Oskar Lange and Eliseo Papa and Gabriella Stanton and Victor Martin and Sukhdeep Singh and Lai H. Wong and Russ Bates and Simon A. Kohl and Josh Abramson and Andrew W. Senior and Yilmaz Alguel and Mary Y. Wu and Irene M. Aspalter and Katie Bentley and David L. V. Bauer and Peter Cherepanov and Demis Hassabis and Pushmeet Kohli and Rob Fergus and Jue Wang},
journal= {arXiv preprint arXiv:2409.08022},
year = {2024}
}
Comments
45 pages, 17 figures