‘New era in digital biology’: AI reveals structures of almost all known proteins | science

What a difference a year makes. Twelve months ago, the artificial intelligence (AI) company DeepMind surprised many scientists with the release of predicted structures for about 350,000 proteins, part of the work recognized as scienceFlash forward to the year 2021. Yesterday, DeepMind and its partners went much, much further. The company unveiled the probable structures of almost every known protein, more than 200 million bacteria to humans, an amazing achievement for AI and a potential treasure trove for drug development and evolutionary studies.

“We are now releasing the structures for the entire protein universe,” Demis Hassabis, founder and CEO of DeepMind, said at a press conference in London.

The structural payoff comes from AlphaFold, one of a number of new AI programs that have solved the protein folding problem, the long-standing challenge of accurately deriving the 3D shapes of proteins from their amino acid sequences. The new predicted structures of AlphaFold were published yesterday in an existing database through a partnership with the European Bioinformatics Institute (EMBL-EBI) of the European Molecular Biology Laboratory. The database “has given structural biologists this powerful new tool where you can look up the 3D structure of a protein almost as easily as you can do a keyword search on Google,” Hassabis said.

Eric Topol, director of the Scripps Research Translational Institute, echoed the surprise of many outside scientists. “AlphaFold is the singular and momentous breakthrough in the life sciences that demonstrates the power of AI,” he tweeted. “With this new addition of structures illuminating almost the entire protein universe, we can expect more biological mysteries to be solved every day.”

The launch of the DeepMind framework is “remarkable”, Ewan Birney, deputy CEO of EMBL, said at the press conference. “It will make many researchers around the world think about what experiments they can do now.”

The proteins resolved by AlphaFold come from organisms ranging from bacteria to plants to vertebrates, including mice, zebrafish and humans. Kathryn Tunyasuvunakool, a research scientist at DeepMind, said AlphaFold took about 10 to 20 seconds to make each protein prediction. The company had to work closely with EMBL-EBI, he noted, to figure out how to present the huge number of structures in the database.

DeepMind says more than 500,000 researchers have already used the database since its launch last year. Hassabis predicted a “new era in digital biology” in which drug developers could move from AI-predicted structures of proteins important for any medical condition to using AI to design small molecules that influence these proteins and therefore treat a disease.

Others use the structure predictions to develop vaccine candidates, probe basic questions in biology, such as how the so-called nuclear pore complex controls which molecules enter a cell’s nucleus, or examine the evolution of proteins when life first evolved.

Hassabis, however, warned that the release of the structures is only a starting point. “Obviously, there’s still a lot of biology and a lot of chemistry that needs to be done.”

Leave a Comment

Your email address will not be published.