The DeepMind AlphaFold algorithm of the AI lab has predicted the structures of more than 200 million proteins. These are almost all compounds known to science found in plants, bacteria and animals.
Today in partnership with @emblebi, we’re releasing predicted structures for nearly all catalogued proteins known to science, which will expand the #AlphaFold database by over 200x – from nearly 1 million to 200+ million structures: https://t.co/GjVES2pBFY 1/ pic.twitter.com/lp8qunbUiX
— DeepMind (@DeepMind) July 28, 2022
According to the developers, thanks to the open source code of AlphaFold, scientists from all over the world can use it in their research. In July 2021, the algorithm was deciphering 350,000 3D structures. Since then, thousands of scientists have used the system, DeepMind said.
“More than 500,000 researchers and biologists have used the database to view more than 2 million structures. And these predictive structures have helped scientists make brilliant new discoveries,” said DeepMind founder and CEO Demis Hassabis.
For example, in April 2022, Yale University scientists used the AlphaFold database to develop a new high-performance malaria vaccines. In July 2021, researchers at the University of Portsmouth turned to the system to create enzymes that would fight single-use plastic pollution.
“This moves us forward a year, if not two,” said John McGeehan, director of the Portsmouth Enzyme Innovation Center.
< p> DeepMind also reported that over the past year, scientists have published over 1000 papers in which AlphaFold was used. In the future, the researchers plan to use the algorithm to create drugs against little-studied but widespread tropical diseases like leishmaniasis.
Recall that in July 2021, DeepMind introduced the AlphaFold algorithm, which simulated 20,000 human protein structures. The developers have opened access to the program for researchers from all over the world.
In November, the Alphabet holding founded Isomorphic Laboratories, which uses artificial intelligence to search for new types of drugs.
In July 2022, researchers from MIT has developed the EquiBind deep learning model, which is 1200 times faster than peers in binding molecules to proteins to create drugs.