DeepMind research cracks the structure of almost every known protein
Alphabet-owned AI company's breakthrough will significantly shorten the time it takes for biological discoveries
Marcus Stuart (MS)May 8, 2025
Artificial intelligence has pushed the boundaries of scientific knowledge by being able to predict the shape of almost every known protein - a breakthrough that will significantly reduce the time it takes for biological discoveries.
The research was conducted by London-based AI firm DeepMind, which is owned by Google parent company Alphabet. She used her AlphaFold algorithm to create the most complete and accurate database of the more than 200 million known proteins to date.
Predicting the structure of a protein from its DNA sequence alone is one of the greatest challenges in biology. Current experimental methods for determining the shape of a single protein take months or years in the laboratory, which is why only about 190,000, or 0.1 percent of the known protein structures have been solved so far.
DeepMind CEO Demis Hassabis said the AI has "given structural biologists this powerful new tool that allows you to look up the 3D structure of a protein almost as easily as a Google search."
"This opens up tremendous opportunities for AlphaFold to make an impact on sustainability, food security and neglected diseases," he added.
In July 2021, DeepMind announced that it had predicted the shape of all human proteins, contributing to a better understanding of human health and disease. This database has been expanded 200-fold and now contains more than 200 million predicted protein structures, covering almost every organism on earth whose genome has been sequenced - from the malaria parasite to the honey bee.
These structures are now available through a public database operated by the European Bioinformatics Institute at the European Molecular Biology Laboratory (EMBL-EBI). In the year that has passed since launch, more than 500,000 researchers around the world have accessed the AlphaFold database to view more than 2 million structures, according to the company.
'Almost every drug that has come to market in recent years has been developed in part using knowledge of protein structures,' said Janet Thornton, a senior scientist and Emeritus Director of EMBL-EBI. "Having access to all these new structures, especially for unusual organisms for which we didn't have structural data, offers a real opportunity not only to develop new drugs, but also to ensure that these drugs do not encounter human proteins and cause cross-reactions."
Proteins are often referred to as the building blocks of life. Their structures are important because they determine how the proteins carry out their tasks. Knowing the shape of a protein, e.g. B. a Y-shaped antibody, gives the scientists information about the role of this protein.
Being able to predict the shape of a protein could allow scientists to control and alter it so that they can improve its function by altering its DNA sequence or target drugs to it. For example, examining the surface proteins of a malaria parasite can help to understand how antibodies bind to the parasite and thus how the pathogen can be effectively combated.
'The use of AlphaFold was really groundbreaking and gave us a sharp look at [a] malaria surface protein,' said Matthew Higgins, professor of biochemistry at the University of Oxford who studies malaria. His team is using these findings to develop a new malaria vaccine, he said.
While scientists still need to confirm a protein's structure through experiments, these predictions give them a huge head start and reduce the time it takes to complete the process.
DeepMind said it excluded viruses from its database to prevent that data from being weaponized by malicious actors or bio-terrorists.
In November 2021, DeepMind announced a spin-off company, Isomorphic Labs, that aims to use AlphaFold and other AI tools to accelerate drug discovery. On Thursday, the company announced that it will open a traditional wet laboratory at the Francis Crick Institute for this purpose.
"We can start thinking about end-to-end drug design. That would be my dream if we could speed up the whole process, not just the structural parts... for new drugs and cures," Hassabis said. "That's coming."
