DeepMind AI successfully anticipates drug-protein interactions

Artificial intelligence systems can now determine not only how proteins fold, but also how they interact with other proteins, drug molecules, or DNA. Biochemists and pharmaceutical researchers say the tool has the potential to greatly speed up tasks such as discovering new drugs.

Proteins, which play many important roles in living organisms, are composed of chains of amino acids, but their complex three-dimensional shapes are difficult to predict.

In 2020, AI company DeepMind first announced that its AlphaFold AI could accurately predict protein structures from amino acid sequences, solving one of biology’s biggest challenges. The company announced that it had mapped 98.5% of the proteins in the human body by mid-2021.

With the latest version of AlphaFold 3, researchers can now model how proteins, including antibodies, not only interact with each other, but also with other biomolecules such as DNA and RNA strands. DeepMind says its predictions are at least 50% more accurate than existing methods.

Most drug molecules work by binding to specific sites on proteins. AlphaFold 3 rapidly speeds the development of new drugs by creating a rapid method to test how candidate drug molecules interact with proteins in silico before running long and expensive clinical tests. There is a possibility that it can be uploaded.

As with previous versions of AlphaFold, the models of proteins or their interactions generated by the latest update have not been experimentally validated. Demis Hassabis, DeepMind’s chief executive, said AlphaFold 3 only provides predictions, so laboratory validation remains important, but that research will be “significantly accelerated” in the future. Masu.

julian bergeron A professor at King’s College London, who was not involved in the development of AlphaFold 3 but has been testing it for several months, says AlphaFold 3 has changed the way experiments are performed. “You can start testing your hypotheses before you even go to the lab. This will be truly transformative. I think every structural biology or protein biochemistry research group in the world will immediately adopt this system. I’m sure,” he says.

keith whison Researchers at Imperial College London say the tool has the potential to streamline much of drug discovery and biology research, allowing researchers to focus on useful molecules they may not have previously discovered. He says it will be possible.

“Organic chemists used to say that chemical space is larger than the number of atoms in the universe, and that we will never be able to access even the most remote and smallest parts of it. But these AI “I think the technology will give us access to a huge amount of relevant chemical space,” he says.

matt higgins from the University of Oxford says DeepMind’s new AI capabilities will revolutionize biomedical researchers, including his own work studying host-parasite interactions in malaria.

“AlphaFold has revolutionized our ability to predict the structure of protein molecules, but the protein machinery used in our cells rarely works in isolation,” he says. “AlphaFold 3 brings new and exciting capabilities to modify protein molecules with the most common additions and to attach them to the most common binding partners in our bodies and see what happens. Masu.”