DeepMind uses AI to predict the structure of almost all proteins

DeepMind, with the European Bioinformatics Institute, has unveiled predicted structures for most known proteins, expanding their AlphaFold Protein Structure Database with over 200 million predictions. This resource aims to propel research in areas like neglected diseases, food insecurity, and sustainability.

DeepMind, in partnership with the European Molecular Biology Laboratory’s European Bioinformatics Institute, has released predicted structures for nearly all catalogued proteins known to science. The announcement comes a year after the two partners released and open-sourced AlphaFold – an artificial intelligence (AI) system used to predict the 3D structure of a protein – and created the AlphaFold Protein Structure Database to share this scientific knowledge with the researchers. The database now contains over 200 million predicted protein structures, covering plants, bacteria, animals, and other organisms. It is expected to help researchers advance work on issues such as neglected diseases, food insecurity, and sustainability.