The interesting moment comes when a designed sequence becomes a physical molecule—and does what its designers hoped.
Designing around a moving target
In a 2025 Nature paper, Liu and colleagues used RFdiffusion to design binders for proteins and regions that do not hold a single fixed shape. Starting from a target sequence, the method sampled the shapes of both target and binder. The team then tested the designs experimentally.
Liu et al. · Nature, 2025Beyond a good fit
The study included binders to amylin, a hormone associated with amyloid deposits in type 2 diabetes. In laboratory experiments, amylin binders inhibited fibril formation and dissociated existing fibres. The structure shown here is an experimentally determined amylin–binder complex. These are research results; they do not establish a treatment for patients.
Amylin–binder structure · PDB 9CC5Where the story goes next
- How many designs were made and tested before a useful binder emerged?
- Does it bind selectively, and does binding change the biology we care about?
- Can it reach its target, remain stable and function in the setting where it would be used?
Editorial preview · Coverage in development