Our latest paper focuses on using weak information to infer protein structures. We use heuristic information—that proteins have hydrophobic cores, secondary structures, and form hydrogen bonds—to accelerate protein folding. The challenge is that we don't know, for example, exactly which residues form hydrogen bonds or how they pair up. However, even this uncertain knowledge is enough to accelerate conformational search by many orders of magnitude.
Structure prediction paper published in PNAS
