The British pharma giant
is teaming up with the U.S. chip maker
to apply artificial intelligence to the discovery of new drugs and vaccines.
The companies said on Monday that their scientists will collaborate at the drug company’s new AI center in London, as well as at Cambridge, where a consortium also announced the building of the U.K.’s most powerful supercomputer, using Nvidia technology.
GSK’s head of AI and machine learning, Kim Branson, said the Nvidia systems will allow his researchers to search for new medicines by poring through vast databases of pathology images and genomic signals.
“The things we’re doing haven’t been done before,” he told Barron’s.
In premarket trading on Monday morning, shares of GlaxoSmithKline (ticker: GSK) gained 0.9%, while stock in Nvidia (NVDA) rose 1.3%. Futures on the
were up 0.7%.
Since GSK started its AI unit in 2019, Branson’s team has grown to one of the largest groups in the drug industry bringing machine learning to the hunt for treatments.
The chips that Nvidia invented to speed up videogames and other image processing have also proven popular for at scientific computing. So the Santa Clara, Calif.-based company has developed specialized computers and software for its scientific customers.
The Nvidia vice president in charge of health care applications, Kimberly Powell, said the company’s systems assist radiologists in examining images, and were used this year at Oak Ridge National Laboratory to scan billions of chemical compounds for potential drugs against Covid-19.
“We call it computational drug discovery,” she said.
GSK’s Branson said that his computer researchers look for links between disease and genetic activity. Then the team goes to the lab to test those associations, with tools that turn genomic machinery up or down. It’s a feedback loop, he says, between data science and laboratory genomics.
When tasks exceed the capabilities of GSK’s in-house systems, Branson says, his researchers will make use of the newly-announced supercomputer in Cambridge.
The Cambridge-1 supercomputer is expected to be the 29th most powerful in the world when it lights up in December. Along with GSK, it will be used by
(AZN), medical researchers at British universities, and AI startups.
Nvidia says it is will be the first system to use its technology in a shared hub, accessible by external researchers. The supercomputer will also use chips from Arm Holdings—the circuit designer that Nvidia agreed to acquire in September from SoftBank, for cash and stock worth $40 billion.
Write to Bill Alpert at email@example.com