Within any living organism, there are thousands of different proteins, each with its own unique shape. For decades, the exact formation of those shapes has been a pain for scientists to figure out. How exactly does a protein, which starts as a string of amino acids, fold itself into the funky 3D shapes you might recognize from diagrams? AlphaFold, an AI from DeepMind, may have an answer. It can predict, with heretofore unseen accuracy, the shape a protein will take.
AlphaFold was put to the test in a global competition called Critical Assessment of protein Structure Prediction, or CASP, which DeepMind CEO Demis Hassabis calls the “Olympics of protein folding” in a video. During the competition, systems like AlphaFold are given the amino acid strings for proteins with shapes that have already been identified through previous experiments but haven’t been published yet. Judges compare the protein shapes produced by the systems with what they know the shapes should be.
At the end of the competition, AlphaFold had the most accurate predictions of any CASP participant in its 25-year history by a wide margin. Even the predictions that weren’t accurate enough to be considered “competitive” with experimental results were only a few atom-widths off. The complete data still needs to be peer-reviewed and published, but the DeepMind team is excited about the results so far, saying in a blog post that they are “optimistic about the impact AlphaFold can have on biological research and the wider world.”
It can take years in the lab for scientists to identify the shapes of individual proteins. Neural networks like AlphaFold could help speed up biological research and drug development in the future. The AI method isn’t perfect yet, and it won’t be taking over for flesh-and-blood researchers anytime soon, but it could be a major step in the everlasting marathon of scientific advancement. To learn more about AlphaFold and to see a bunch of scientists and engineers getting giddy and fist-pumping over competition results.