Paper
De novo design of high-affinity protein binders with AlphaProteo
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.
Authors: Zambaldi, Vinicius · La, David · Chu, Alexander E. · Patani, Harshnira · Danson, Amy E. · Kwan, Tristan O. C. · Frerix, Thomas · Schneider, Rosalia G. · Saxton, David · Thillaisundaram, Ashok · Wu, Zachary · Moraes, Isabel · Lange, Oskar · Papa, Eliseo · Stanton, Gabriella · Martin, Victor · Singh, Sukhdeep · Wong, Lai H. · Bates, Russ · Kohl, Simon A. · Abramson, Josh · Senior, Andrew W. · Alguel, Yilmaz · Wu, Mary Y. · Aspalter, Irene M. · Bentley, Katie · Bauer, David L. V. · Cherepanov, Peter · Hassabis, Demis · Kohli, Pushmeet · Fergus, Rob · Wang, Jue