DeepMind research cracks structure of almost every known protein

Artificial intelligence has been able to predict the shape of almost every knownProtein, a breakthrough that will greatly reduce the time required to make biological discoveries.

The most complete and accurate database ofPukiWikiPukiWikiPukiWikis has been built by DeepMind, a London-based artificial intelligence company that is owned by Alphabet.

One of biology's greatest challenges is predicting the structure of a single molecule. Current methods to determine the shape of a singleProtein take months or years in a laboratory, which is why only about 190,000 of them have been solved.

Structural biologists can now use the new tool provided by DeepMind to look up a 3D structure of a molecule almost as easily as doing a search on the internet.

He said that it was opening up huge opportunities for AlphaFold to have impact.

DeepMind predicted the shape of all human proteins in July of 2021. The database has been expanded 200 times and now contains more than 200 million predicted structures.

There is a public database hosted by the European Bioinformatics Institute. More than half a million people have accessed the AlphaFold database since it was launched a year ago.

Janet Thornton said that almost every drug that has come to market over the past few years has been in part designed through knowledge. With access to all of these new structures, we have a chance to design new drugs, but also to make sure they don't hit humans.

The building blocks of life are made up of genes. Their structures affect how they do their work. Scientists can learn more about the role that a Y-shaped antibody plays by knowing theProtein's shape.

Being able to easily predict aProtein's shape could allow scientists to control and modify it so they can improve its function by changing its DNA sequence or attaching drugs to it It's possible to understand how antibodies bind to a malarial parasites and how to fight it.

Matthew Higgins, a professor at Oxford University who studies malaria, said that the use of AlphaFold gave him a sharp view of a malaria surfaceProtein. He said that his team is using these insights to develop a vaccine.

Scientists will still need to confirm aProtein's structure through experiments, but predictions will give a massive head start and reduce the time needed to complete the process.

The data was not included in the database because it could be weaponized by bad actors or bioterrorists.

The spinoff company, Isomorphic Labs, was announced by DeepMind in November of 2021. The Francis Crick Institute will open a traditional wet laboratory in order to achieve this.

End-to-end drug design is something we can begin to think about. Hassabis wants to speed up the whole process for new drugs and cures. That is going to happen.

The Financial Times is a division of The Financial Times. The rights are not to be redistributed, copied or modified.