A new database includes the predicted shape of nearly every protein found in the human body, BBC News reports, adding that this catalog could help speed drug development.
Researchers from DeepMind, a subsidiary of Alphabet, and their colleagues at the European Molecular Biology Laboratory used the AlphaFold machine-learning program to determine the structure of 98.5 percent of the proteins of the human proteome. As they report in the journal Nature this week, the DeepMind researchers have confidently predicted the structural positions of about 58 percent of amino acids and have a very high degree of confidence for about 36 percent of amino acids.
The researchers are making this data as well as protein structure predictions for 20 model organisms — a total of about 350,000 protein structures — available under a Creative Commons license through the European Bioinformatics Institute, Ars Technica adds.
"Structural biologists are not yet used to the idea that they can just look up anything in a matter of seconds, rather than take years to experimentally determine these things," Demis Hassabis, DeepMind founder and CEO, says at TechCrunch. "And I think that should lead to whole new types of approaches to questions that can be asked and experiments that can be done."
The BBC notes that the database could help in developing new drugs and TechCrunch adds that DeepMind and EMBL have partnered with the Drugs for Neglected Diseases Institute.