The AI-powered software, programmed by David Baker’s group at the Institute for Protein Design in Seattle, is the latest advance in what’s becoming an increasingly heated race by companies and research labs to build on already cutting-edge artificial intelligence technologies used for protein design.
They expand on two earlier tools from Baker’s lab: RoseTTAFold, which uses deep neural networks to predict protein structure; and RFdiffusion, which uses generative AI to create new protein structures from scratch. Those older tools, as well as AlphaFold, the popular protein-prediction software developed by Google’s DeepMind, focus largely on the amino acid building blocks of proteins but weren’t designed to capture how proteins interact with other biomolecules.