The DeepMind database included almost every protein known to science. About 200 million

Last year, DeepMind released an open-source database containing 3D structures of hundreds of thousands

Now, the AlphaFold Protein Structure Database has been expanded to 200 million structures, including nearly every protein known to science.

Proteins are made up of chains of amino acids thatfold into intricate three-dimensional shapes that determine their function. Mapping the structures of proteins is important for understanding what they do and how they work and how things can go wrong. However, it is still difficult to calculate the exact structure of a protein based on its constituent amino acids. This usually requires a huge amount of computing power and man-hours.

This was the case until Alphabethas directed its powerful artificial intelligence DeepMind to solve this problem. Initially trained on 100,000 known protein structures, the system has developed the ability to predict the structures of many millions of other proteins, each taking minutes or seconds to determine, rather than months or years.

Recently, DeepMind released a new large-scaleupdating the database, which now includes about 214 million structures from a million species. This covers almost every protein known to science, which is a huge help for research into treatments for diseases, vaccines, antibiotic resistance, and even plastic pollution.

The entire protein structure database of over 25 terabytes of data can be downloaded from Google Cloud Public Datasets.