DeepMind's protein-folding AI cracks biology's biggest problem

One of the grand challenges of biology has been solved thanks to an artificial intelligence called AlphaFold. The work could speed up the discovery of new drugs because it has led to advances in fighting Malaria, antibiotic resistance and plastic waste.

It has been a problem in biology for a long time to determine the crumpled shapes of proteins based on their sequence of Amino Acids. Some are attracted to others, some are repelled by water, and some form intricate shapes that are hard to determine.

DeepMind first announced it had developed a method to accurately predict the structure of folded proteins in late 2020, and by the middle of 2021.

The company announced today that it is publishing the structures of more than 200 millionProteins, nearly all of which are catalogued on the globally recognised repository ofProtein research, UniProt

The European Bioinformatics Institute has collaborated with DeepMind to create a database that can be easily accessed by researchers around the world. The AlphaFold protein structure database is a gift to humanity.

He says that he has seen many of these moments where you can sense the landscape changing under you and the provision of new resources. Two years ago, we didn't know that this was doable.

Already delivering results

The CEO of DeepMind says that it's almost as easy to find aProtein structure using the database as it is to do a search on the internet. The company is owned by the parent company.

Scientists use the archive in a number of areas. The researchers at the University of Oxford were trying to map the structure of aProtein that they believed was key to disrupting the life cycle of the Malaria parasites.

X-ray crystallography is one of the methods used. To see what the molecule looks like, we fire X-rays at it, and then use the X-ray data to see what the molecule looks like. We were never able to see in enough detail what this molecule looked like.

The nature of life is one of the toughest problems in science, according to DeepMind's co- founder.

When AlphaFold was released, it gave a prediction of the structure that was similar to what the researchers had found. They hope that they can design a vaccine that will work as a malarial vaccine.

X-ray crystallography is expensive and takes a lot of time. AlphaFold hasn't had to make choices about what they do. "I think we can be confident that there are new experiments and new insights coming through due to AlphaFold, which will impact 'how does this particular parasites work' or 'why does this particular disease happen in humans'."

AlphaFold can be used to engineer new enzymes to break down plastic waste, and to learn more about the genes that makebacteria resistant to antibiotics.

Work still to be done

AlphaFold has changed the world of biological research, but there are still problems to be solved in the field.

It was great when AlphaFold came out. He says that you just take your favourite proteins and look them up. It took me eight years to do the crystallographic structure. People are making fun of the fact that crystallographers will be out of work.

Pushmeet Kohli is a researcher at DeepMind.

AlphaFold doesn't have the ability to model exactly how they fold. Instead, it is only able to use some of the parts of the proteins that have been shown to work.

The structures of the tool are predictions rather than calculated results. AlphaFold has yet to solve the complex interactions between proteins or even make a difference in a small subset of structures that seem to have unpredictable folding patterns.

There are more issues thrown up after you discover one thing. How complicated biology is really frightening.

Tomek Wlodarski at University College London says that AlphaFold has had an enormous impact on many areas of biology but that there are improvements to be made on accuracy.

AlphaFold indicates which parts of a prediction have a high accuracy and which parts are less confident in.

Artificial intelligence that can reason and apply knowledge is a goal of many researchers. They should be looking at psychology as well.

He says that AlphaFold gave him the same structure as it gave without the Mutation. I did another test to see if I could remove the residues from one of the ends of the proteins. I was able to cut half of the sequence, and the program still predicted it to be a folded version of the same structure. There are some issues.

DeepMind is still working on improving the accuracy and capabilities of AlphaFold, according to Pushmeet Kohli, who leads the scientific team.

He says that the game isn't over even though he knows the static structure of proteins. We want to understand how the proteins interact with one another. We want to understand how the recipe of life is translated into which genes are created, when they are created, and how a cell works.

You can get the health, diet and fitness news you need every Saturday by signing up for our Health Check newsletter.