Details of RFpeptides released

University of Washington researchers have lifted the lid on a tool for artificial intelligence–assisted macrocycle design licensed to Vilya, one of C&EN’s 10 Start-Ups to Watch for 2024 (bioRxiv 2024, DOI: 10.1101/2024.11.18.622547). The work has been published as a preprint and has not yet been peer reviewed.

The paper describes the program RFpeptides, which was developed in the labs of Gaurav Bhardwaj, Frank DiMaio, and 2024 Nobel laureate David Baker. The tool builds on protein structure prediction software RoseTTAFold 2 and its sibling RFdiffusion, which generates new protein structures.

RFpeptides creates macrocyclic peptides designed to bind to a specified target, and it doesn’t need to build off an existing peptide structure. Cyrus Harmon, CEO of Vilya, previously told C&EN that these rings are stabler than linear peptides, which are more “floppy” and prone to getting degraded in the body. They could be a midpoint between large biologics, which can bind to a large surface but are too big to be orally bioavailable, and small molecules, which can be ingested but are too small to bind to large surface areas.