A year after demonstrating its AlphaFold program could tackle the decades-old problem of predicting the final working shapes of proteins from just their building blocks—and making it open to the public—DeepMind is now expanding to nearly every protein known to science.
The new details will help researchers visualize the nearly 200 million proteins that form the basis of life for animals, plants, bacteria and more—spanning nearly every organism on the planet that has had its genome sequenced.
It’s a major evolution of the artificial intelligence system that previously cataloged the intricate structures of nearly 1 million proteins from just 10,000 species, with initial focuses on research areas such as neglected and tropical diseases.
“By demonstrating that AI could accurately predict the shape of a protein down to atomic accuracy, at scale and in minutes, AlphaFold not only provided a solution to a 50-year grand challenge, it also became the first big proof point of our founding thesis: that artificial intelligence can dramatically accelerate scientific discovery, and in turn advance humanity,” said Demis Hassabis, CEO of DeepMind, which falls under the umbrella of Google’s corporate parent Alphabet.
In practice, that means that most pages on the international protein database UniProt will now come with 3D models—many out of reach until now—in addition to their listings of amino acids and genetic sequences. That will bring researchers a big step closer to finding ways to manipulate them, not only to help combat diseases, but to address issues such as plastic pollution and food insecurity, according to DeepMind.
“Determining the 3D structure of a protein used to take many months or years, it now takes seconds,” said Eric Topol, director of the Scripps Research Translational Institute. “And with this new addition of structures illuminating nearly the entire protein universe, we can expect more biological mysteries to be solved each day.”
DeepMind first launched the public AlphaFold database in July 2021 in partnership with the European Bioinformatics Institute, part of the intergovernmental European Molecular Biology Laboratory, starting with only about 350,000 protein structures and spanning the entire human proteome.
Over the following 12 months, it has grown in size to support more than 500,000 researchers from over 190 countries.
“In the past year alone, there have been over a thousand scientific articles on a broad range of research topics which use AlphaFold structures; I have never seen anything like it,” said Sameer Velankar, team leader at EMBL-EBI’s Protein Data Bank in Europe. “And this is just the impact of one million predictions; imagine the impact of having over 200 million protein structure predictions openly accessible in the AlphaFold Database.”
AlphaFold has already shown its worth in drug discovery, with biotechs importing its predictions into their own computer models to explore how proteins will interact with potential medicines.
"AlphaFold became an essential tool for biopharma research nearly overnight, including here at Rome Therapeutics where it is allowing us to predict protein structures in areas of the dark genome that have never been solved for before,” said Rosana Kapeller, CEO of the former Fierce 15 winner which launched in 2020.
It’s also helped drug designers be more accurate and avoid potential side effects, by not only offering predictions on the proteins that will connect with a therapy, but also those that will not.
“When you come across a target for a new drug, you often find that the receptors—proteins that bind to the drug—are part of a family. Looking for one target means coming across other targets that are like brothers and sisters and cousins,” Karen Akinsanya, president of R&D at Schrödinger, wrote in a DeepMind blog post.
“The challenge for people working in drug discovery is finding a drug or molecule that binds one member of that family and inhibits that family member but doesn’t inhibit the rest of the family. In part, this is where AlphaFold has worked so brilliantly for us,” Akinsanya said.