Speaking at a press briefing Tuesday, DeepMind's cofounder and CEO, Demis Hassabis, said the expanded database effectively covered 'the entire protein universe' and would make it as easy to look up a 3D protein structure as typing out 'a keyword Google search.'

The firm's researchers, working in partnership with the European Molecular Biology Laboratory, have spent the past year using AlphaFold to expand the firm's database from 1 million protein structures to more than 200 million and making them freely available.

The open-source database could aid new scientific discoveries, the company said.Īlphabet's artificial-intelligence research unit, DeepMind, has predicted a 'new wave of scientific discoveries' after unveiling a trove of 200 million free-to-access models of microscopic protein structures.ĭeepMind, which began life as a startup in London before Google (now Alphabet) bought it in 2014, says it has used its AI program AlphaFold to predict the 3D structures for almost all catalogued proteins known to science.

Proteins are a key building block for life, but their structures are difficult to predict and study.

DeepMind has used AI to predict 200 million protein structures and is making them freely available.

Typing bolt free download