Design / Generative protein design
Proteins we can imagine
Generative models propose new protein structures. Experiments determine which designs actually work.
What the evidence shows
RFdiffusion demonstrated de novo protein design with experimental validation for selected designs.
What we can reasonably infer
Design tools may make specialized binders and industrial proteins easier to explore when paired with testing.
What remains a possibility
Designing arbitrary biological functions on demand, with predictable outcomes, is a much larger goal.
THE BOUNDARY
Where the claim stops
Computational success and laboratory success are different outcomes. Therapeutic use needs additional validation.
THE NEXT USEFUL SIGNAL
What to watch for
Reproducible design success on new tasks, compared with conventional methods.
The horizon is an editorial scenario for this brief’s central possibility. It does not predict a treatment date or an investment return.