This article is more than 1 year old

AI-predicted protein structures could unlock vaccine for COVID-19 coronavirus... if correct... after clinical trials

It's not quite DeepMind's 'Come with me if you want to live' moment, but it's close, maybe

DeepMind has shared its AI software's homework detailing the structure of six proteins linked to SARS-CoV-2 aka the coronavirus that causes COVID-19.

The machine-learning startup hopes the findings can help scientists develop a vaccine to halt and reverse the global spread of the bio-nasty. Knowing these viral protein structures, boffins can grasp how the virus works, and how it can be stopped. Some labs have shared their own protein data, produced computationally or experimentally, and now DeepMind has waded in with its AI-determined efforts.

The cynical among us will see this as a marketing stunt. Yet, the findings are there to download if you want to give them a try. Pushing that aside, one good reason for using artificial intelligence here is that it is faster than physical laboratory work, and hopefully as accurate.

The upstart's eggheads used AlphaFold, a model unveiled in late 2018, to predict six of the underlying protein structures for SARS-CoV-2. They used data from the Universal Protein Resource, an open database describing the form and function of hundreds of thousands of proteins taken from genome sequencing projects, to aid AlphaFold in its predictions. The software, for what it's worth, uses a technique known as “free modelling” to predict the protein structures of the novel coronavirus.


DeepMind founder behind NHS data slurp to be beamed up to Google mothership


“We emphasise that these structure predictions have not been experimentally verified, but hope they may contribute to the scientific community’s interrogation of how the virus functions, and serve as a hypothesis generation platform for future experimental work in developing therapeutics,” DeepMind said on Thursday.

It’s unclear how accurate any of the six predictions really are since so little is known about the novel coronavirus. One of the models, however, is similar to one experimentally derived by a group of scientists at the University of Texas at Austin and the National Institute of Allergy and Infectious Diseases, an agency under the US National Institute of Health. They both feature a “spike protein structure.”

DeepMind has shared its results with researchers at the Francis Crick Institute, a biomedical research lab in the UK, as well as offering it for download from its website.

“Normally we’d wait to publish this work until it had been peer-reviewed for an academic journal. However, given the potential seriousness and time-sensitivity of the situation, we’re releasing the predicted structures as we have them now, under an open license so that anyone can make use of them,” it said.

There are 93,090 cases of COVID-19, and 3,198 deaths, spread across 76 countries, according to the latest report from the World Health Organization at time of writing. ®

More about


Send us news

Other stories you might like