DeepMind: its AI could predict the structure of all known proteins

In December 2020, DeepMind surprised the world of biology when it solved a grand 50-year challenge with AlphaFold, an AI that predicts the structure of proteins. Last week, the London-based company released full details of the tool and released its source code. The company recently announced that it has used its AI to predict the shapes of almost every protein in the human body, as well as the shapes of hundreds of thousands of other proteins found in 20 of the most studied organisms, including yeast. , fruit flies and mice.

This breakthrough could allow biologists around the world to better understand diseases and develop new drugs. So far, the database consists of 350,000 newly predicted protein structures. DeepMind is due to predict and publish the structures of over 100 million additional proteins over the next few months – roughly all the proteins known to science.

The folding of proteins has been an issue that I have been monitoring for over 20 years. It has been a huge project for us. I would say this is our biggest accomplishment so far. And this is the most exciting in a way, because it is expected to have the biggest impact in the world outside of AI. Says Demis Hassabis, co-founder and CEO of DeepMind.

An essential prediction tool for research

Proteins are made up of long ribbons of amino acids, which twist into complicated knots. Knowing the shape of a protein’s knot can reveal what that protein does, which is crucial for understanding how diseases work and developing new drugs, or identifying organisms that can help fight pollution and climate change. Determining the shape of a protein takes weeks or months in the lab. AlphaFold can predict shapes to the nearest atom in a day or two.

The new database should make life even easier for biologists. AlphaFold may be available to researchers, but not everyone will be able to run the software itself. ” It is much easier to retrieve a structure from the database than to run it on your own computer Explains David Baker of the Institute for Protein Design at the University of Washington, whose laboratory has developed its own tool to predict the structure of proteins, called RoseTTAFold, based on the AlphaFold approach.

AlphaFold can predict the structure of many proteins, helping researchers better understand them and develop more effective therapeutic molecules. © DeepMind

Over the past several months, Baker’s team have worked with biologists who were previously stuck trying to figure out the shape of the proteins they were studying. ” There is a lot of pretty interesting biological research that has been really accelerated. Says Baker. A public database of hundreds of thousands of ready-made protein forms should be an even bigger accelerator. It sounds surprisingly impressive, according to Tom Ellis, a synthetic biologist at Imperial College London. But he cautions that most of the predicted forms have yet to be verified in the lab.

Precision on an atomic scale

In the new version of AlphaFold, the predictions come with a confidence score that the tool uses to indicate how close it considers each predicted shape to be. Using this measurement, DeepMind found that AlphaFold predicted the shapes of 36% of human proteins with correct accuracy down to the level of individual atoms. It’s precise enough for drug development, Hassabis says. Previously, after decades of work, only 17% of the proteins in the human body had had their structures identified in the laboratory.

If AlphaFold’s predictions are as accurate as DeepMind claims, the tool would have more than doubled that number in just a few weeks. Even predictions that are not entirely accurate at the atomic level are still useful. For more than half of the proteins in the human body, AlphaFold predicted a shape that should be good enough for researchers to understand the protein’s function. The rest of AlphaFold’s current predictions are either incorrect or concern the one-third of the proteins in the human body that have no structure until they bind to each other.

The fact that it can be applied at this level of quality is impressive Says Mohammed AlQuraish, a systems biologist at Columbia University, who has developed his own software to predict the structure of proteins. He also points out that having structures for most of the proteins in an organism will allow us to study how these proteins function as a system, and not just in isolation. ” This is what seems most exciting to me “.

Sources: Nature

Back to top button