RFdiffusion scaffolds help solve four small-protein structures

AI system

System: RFdiffusion (generative protein backbone diffusion model) with ProteinMPNN sequence design
Model: RFdiffusion designs protein backbones, and ProteinMPNN designs their amino-acid sequences.
Release: Exact release not established.
Application developer: University of Washington Institute for Protein Design (Baker lab)
Role: RFdiffusion generated protein backbones, and ProteinMPNN generated amino-acid sequences.
Configuration: Not assessed
Inputs and outputs: Not assessed
Autonomy: Not assessed
Workflow: Publicly released RFdiffusion and ProteinMPNN models. A nanobody against each target and a rigid trimbody scaffold design. Cryo-EM data collection and processing by the deploying group.
Human oversight: Not assessed
Relation to AI: RFdiffusion and ProteinMPNN produced the de novo scaffold domains that make the assembly rigid. The deploying authors are not the tool's developers. They validated the designs experimentally by determining cryo-EM structures at atomic resolution. The designs originated with the deploying group, which used the publicly released model.
AI contribution: RFdiffusion and ProteinMPNN produced the de novo scaffold domains that make the assembly rigid. The deploying authors are not the tool's developers. They validated the designs experimentally by determining cryo-EM structures at atomic resolution. The designs originated with the deploying group, which used the publicly released model.

Studies: Not assessed
Study design: Not assessed
Measurement: Not assessed
Setting and sample: Not assessed
Measures: four small-protein structures at 2.29 to 2.62 angstroms
Comparison: earlier DARPin-containing cage fiducials for small proteins
Results: Reference: determined the structure of the cargo protein (GFP) at a local resolution of 3.8 Å but failed to provide a reliable atomic interpretation. Result: Aequorea coerulescens GFP (2.29 Å).
Additional value from AI: Not assessed
Limitations: Not assessed
Denominators: Not assessed
Materials for replication: Not assessed

Participants: Developer: University of Washington Institute for Protein Design (Baker lab). Operator: Shandong University and collaborating institutes.
Jurisdictions: Benefit: Jinan, China. Developer: United States. Operator: China.
Ownership and control: Not assessed
Dependencies: Not assessed
Cross-border connections: Not assessed
Geopolitical implications: Not assessed

Conditions for replication: Publicly released RFdiffusion and ProteinMPNN weights. A nanobody against the target protein. Cryo-EM data collection and processing capability.
Replication attempts: Not assessed
Later findings: No later development found.
Availability: Not assessed
Institutional decision: Not assessed
Grounds: Not assessed
Action and result: Not assessed
Success criterion: Not assessed

Assessed benefits: Organizations. Society.
Depth: Organizations: 2. Society: 1.
Reasons: Organizations: The deploying group obtained a reusable fiducial. It yielded four structures between 2.29 and 2.62 angstroms, for proteins that are otherwise difficult targets for cryo-EM. Society: A methodological advance for structural biology with no identified downstream beneficiaries reported in the source.
Observed or intended: Not assessed
Rubric: Not assessed
Assessor and date: Not assessed
Alternative assessments: Not assessed
Fill: The deployer's account

Kind: Not assessed
Related entries: Not assessed
Field: Structural biology / de novo protein design.
Purpose: Generate de novo protein backbones to order, here to build rigid scaffolds that hold a small target protein still for cryo-electron microscopy.
Dates: Date basis unassessed: 2026 (precision unassessed). Updated: 21 October 2026.
Stage: Not assessed
Version: SHA-256: 732742459932…