Protein Design Using Structure-Prediction Networks: AlphaFold and RoseTTAFold as Protein Structure Foundation Models

Jue Wang; Joseph L Watson; Sidney L Lisanza

doi:10.1101/cshperspect.a041472

Protein Design Using Structure-Prediction Networks: AlphaFold and RoseTTAFold as Protein Structure Foundation Models

Cold Spring Harb Perspect Biol. 2024 Mar 4:a041472. doi: 10.1101/cshperspect.a041472. Online ahead of print.

Authors

Jue Wang^{1

2

3

4}, Joseph L Watson^{5

2}, Sidney L Lisanza^{5

2

3}

Affiliations

¹ Department of Biochemistry, University of Washington, Seattle, Washington 98195, USA juewang@post.harvard.edu.
² Institute for Protein Design, University of Washington, Seattle, Washington 98195, USA.
³ Graduate Program in Biological Physics, Structure and Design, University of Washington, Seattle, Washington 98195, USA.
⁴ DeepMind, London EC4A 3BF, United Kingdom.
⁵ Department of Biochemistry, University of Washington, Seattle, Washington 98195, USA.

PMID: 38438190
DOI: 10.1101/cshperspect.a041472

Abstract

Designing proteins with tailored structures and functions is a long-standing goal in bioengineering. Recently, deep learning advances have enabled protein structure prediction at near-experimental accuracy, which has catalyzed progress in protein design as well. We review recent studies that use structure-prediction neural networks to design proteins, via approaches such as activation maximization, inpainting, or denoising diffusion. These methods have led to major improvements over previous methods in wet-lab success rates for designing protein binders, metalloproteins, enzymes, and oligomeric assemblies. These results show that structure-prediction models are a powerful foundation for developing protein-design tools and suggest that continued improvement of their accuracy and generality will be key to unlocking the full potential of protein design.