DeepMind’s artificial intelligence predicted the 3D structures of the “entire protein universe”
Protein structures began to be discovered in the middle of the last century, but for more than 50 years, mankind has not made significant progress in this area. In 1957, biochemist John Kendrew first determined the structure of the protein myoglobin, which helps oxygenate muscles. No matter how groundbreaking the discovery was, over the following decades, scientists identified only about a dozen protein structures.
Now with the help of AI, everything can change. On Thursday, Google subsidiary DeepMind said it had successfully used artificial intelligence to predict the 3D structures of almost all proteins known to science. These are more than 200 million proteins found in plants, bacteria, animals, and humans – in almost all living things that exist on Earth.
AlphaFold, an artificial intelligence system from DeepMind, was able to decipher the protein. It has an open-source database, so scientists from all over the world can use it for their research for free. The program was officially launched in July last year. Then she deciphered about 350,000 3D structures. Since then, scientists have used it many times.
“More than 500,000 researchers and biologists have used the database to view over 2 million structures,” says DeepMind’s founder and CEO.
For example, the database helped Yale University scientists create a new highly effective malaria vaccine. Scientists at the University of Portsmouth used it to develop enzymes that will fight single-use plastic pollution. Other studies include Parkinson’s disease, honey bee health and even archaeology.
Last year alone, scientists published more than 1,000 papers involving AlphaFold. In the coming years, DeepMind also plans to collaborate with an initiative looking for drugs against little-studied but common tropical diseases like leishmaniasis.