AI can predict interactions between proteins and human cells

Google AlphaFold 3

11. May 2024 1:27 p.m. Robert Klatt

AlphaFold 3, a new AI from Google DeepMind, can predict the behavior of proteins in the human body. This is a major advance in medical research with the potential to improve disease treatment and vaccine development.

London, England). There is a scientific theory that the body’s biological processes, including diseases such as cancer, can be better understood by knowing the structure and behavior of their proteins. Google DeepMind, an artificial intelligence (AI) research company, achieved a breakthrough in protein structure prediction in 2020. A large number of researchers have now used AlphaFold 2 AI to make progress significant in medicine, particularly in the development of new vaccines and drugs and in treatments against cancer.

technology/ai/google-deepmind-isomorphic-alphafold-3-ai-model/” target=”_blank”>deep mind and its subsidiary Isomorphic Labs announced a new version of AI. According to the publication in the specialist magazine Nature AlphaFold 3 can predict the interaction of all proteins with other biomolecules present in cells in the human body. AlphaFold 2, on the other hand, was “only” able to predict three-dimensional protein structures, but could not provide any information about how these interact with other proteins.

“In interactions of proteins with other types of molecules, we see at least a 50 percent improvement over existing prediction methods, and in some important categories of interactions we have doubled the prediction accuracy .”

Free tool for scientists

Google DeepMind also announced a free tool for non-commercial researchers. These should be able to use the so-called AlphaFold server to create protein models. This means that doctors and biologists can create complex protein structures on a computer in a short time. In addition, has Isomorphic laboratories explains that she is working with various pharmaceutical companies to use all the possibilities of AI in drug development.

Nature, do I: 10.1038/s41586-024-07487-w