Tag
This paper explores using large language models to generate ranking policies for shortlisting protein binders from candidate pools, showing modest improvements over baseline methods in de novo design workflows.
Anthropic AI announced that Claude successfully designed novel protein binders for 14 out of 15 drug targets, with independent verification by Adaptyv Bio and Twist Bioscience.
This bioRxiv preprint introduces Diff-Switch, a framework that uses diffusion-based ensemble sampling to generate conformational states for de novo protein switch design, improving the success rate of finding switch-compatible sequences.