DeepMind has cataloged almost every protein known to science

Using the AlphaFold Protein Structure Database program, first developed by Alphabet’s DeepMind division in 2018 and released publicly in July 2021, an open-source database of 3D structures of more than 200 million proteins found in plants, bacteria, animals, and humans has been published.

All known protein structures published

Initially trained on 100,000 known protein structures, the system’s ability to predict the structures of millions of other proteins was developed. In July last year, the first AlphaFold Protein Structure Database was made public, containing a total of more than 350,000 protein structures, including approximately 98% of human proteins.

In a new post, the DeepMind team announced that they have updated the AlphaFold Protein Structure Database to include approximately 214 million structures, as a result of their work in partnership with the European Bioinformatics Institute. According to the post, the database now covers almost every protein known to science.

To date, more than 500,000 scientists in 190 countries have accessed the database for use in their research. The database will provide a new resource for studies on topics such as disease treatments, vaccines, antibiotic resistance and even plastic pollution.

More than 200 million builds of over 25 terabytes of data will all be available for bulk download via Google Cloud Public Datasets.