There are thousands of different proteins in every living organism, each with its own unique shape. For decades, the exact formation of these forms has been a pain for scientists to understand. How exactly does a protein that starts as a string of amino acids fold into funky 3D shapes that you can recognize from the charts? AlphaFold, AI from DeepMind, may have an answer. It can predict, with unprecedented accuracy, the shape that the protein will take.
AlphaFold has been tested in a global competition called Critical Assessment of Protein Structure Prediction, or CASP, which DeepMind CEO Demis Hasabis calls the “Protein Folding Olympics”
At the end of the race, AlphaFold had the most accurate predictions from any CASP participant in its 25-year history by a wide margin. Even predictions that were not accurate enough to be considered “competitive” with experimental results were only a few latitudes away from the atom. Full data has yet to be 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 could 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 such as AlphaFold can help accelerate biological research and drug development in the future. The AI method is not yet perfect and will soon take on the work of researchers of flesh and blood, but it could be a major step in the eternal marathon of scientific progress. To learn more about AlphaFold and see a bunch of scientists and engineers getting involved and reminding you of the results of the race, watch the video below.