Biology · 11 July 2023
Proteins designed from a molecular specification, then made
RFdiffusion generated new protein backbones and functional proteins — binders, symmetric assemblies and metal-binding sites — that were experimentally characterized, including a designed influenza binder whose structure nearly matched the design model.
- Field
- Biology
- Date
- 11 July 2023
- AI system
- RFdiffusion
- Institution
- University of Washington, Institute for Protein Design
- Published in
- Nature
- Identifier
- 10.1038/s41586-023-06415-8
RFdiffusion, from David Baker’s group at the University of Washington, is a generative diffusion model built by fine-tuning the RoseTTAFold structure network to denoise protein backbones. Published in Nature on 11 July 2023, it designs proteins from relatively simple molecular specifications: a desired fold, a binding target, a functional motif, or a symmetry.
The authors experimentally characterized hundreds of designed symmetric assemblies, metal-binding proteins and protein binders. In a stringent structural test, they report a cryogenic electron microscopy structure of a designed binder in complex with influenza haemagglutinin that is nearly identical to the computational design model.
This is protein engineering, but the designer is a generative model: given a target or a motif, it proposes new structures that, after sequence design, can be made and tested. The paper’s claim is that a single framework can now solve a wide range of previously difficult design problems, from de novo binders to symmetric nanomaterials, from a brief molecular brief.