3 ms·
Google DeepMind's Demis Hassabis on His Nobel Prize
- dekhn 2y agoI wish he wouldn't say they solved protein folding. No, they greatly improved the accuracy of protein structure prediction (limited mainly to static structures with little to no flexibility, and some ongoing discussion about how well they generalize outside the existing folds of the PDB). "Solving protein folding" is using our understanding of physics to demonstrate we can simulate recapitulate the trajectories followed by natural proteins. That's a different, more challenging, and more fundamental problem. (this is an ongoing discussion in the community, with fewer and fewer people saying "solve protein folding" these days)