A team of researchers from the University of Virginia, including Phil Bourne, Cam Mura, and Eli Draizen, developed an AI-driven approach to explore protein structure relationships. Their study challenges traditional notions and identifies faint relationships missed by traditional methods. The computational framework, called DeepUrfold, can detect and quantify protein relationships at scale. By viewing protein relationships as “communities” and avoiding classification into separate bins, this approach provides a more integrated view. The team detected distant relationships between proteins using DeepUrfold, pushing researchers to move beyond static geometric terms. The study, published in Nature Communications, represents years of work in developing this AI-driven framework.
Source link