News Hub

DeepMind deploys deep learning to predict coronavirus protein structures

Written by Thu 5 Mar 2020

DeepMind hopes predicted protein structures will help scientists develop coronavirus treatments

London-based AI lab DeepMind has published predicted structures for several proteins present in SARS-Cov-2, the virus that causes COVID-19.

DeepMind said that while the predictions have not been experimentally verified, it hopes they will help scientists understand how the virus functions and serve as a platform for developing potential treatments.

The company, which is owned by Google, used the latest version of its AlphaFold deep learning system to predict the structures.

AlphaFold, detailed by DeepMind researchers in a Nature article published in January, is one of a number of computational systems that use algorithms based on “template modelling” to predict the physical structure of proteins from amino acid sequences.

Using a technique called “free modelling”, DeepMind’s own system can predict accurate protein structures even if similar proteins’ structures do not exist.

“Knowing a protein’s structure provides an important resource for understanding how it functions, but experiments to determine the structure can take months or longer, and some prove to be intractable,” company researchers wrote in a blog post. “For this reason, researchers have been developing computational methods to predict protein structure from the amino acid sequence.”

“We’ve continued to improve these methods since that publication and want to provide the most useful predictions, so we’re sharing predicted structures for some of the proteins in SARS-CoV-2 generated using our newly-developed methods,” they added.

While other labs have shared predicted structures of some of COVID-19’s associated proteins, DeepMind said the structures it released relate to under-studied proteins. The company said its own research wouldn’t have been possible without the efforts of other researchers across the globe.

Although AlphaFold is more accurate than its predecessor, DeepMind said the prediction system was “still in development” and cautioned that it could not “be certain of the accuracy of the structures”. The research lab decided to bypass the peer-review process anyway due to the seriousness and urgency of the situation.

Written by Thu 5 Mar 2020


Coronavirus healthcare medicine
Send us a correction Send us a news tip