Oak Ridge National Lab Taps Supercomputers to Fight Coronavirus
Researchers received emergency computation time to run through a database of drug compounds to see which combinations might prevent COVID-19.
As researchers race to identify and unleash new scientific breakthroughs to combat the COVID-19 outbreak, the Energy Department’s Summit supercomputer is playing a role in the fight.
Oak Ridge National Laboratory, where Summit is housed, recently granted researchers emergency computation time to run through a database of existing drug compounds to see which combinations might prevent cell infection of COVID-19. The research is still ongoing, but with the help of Summit, scientists were already able to perform simulations that resulted in outputs that they believe will help pave the way for new, necessary experimentation to support researchers on their quest for a cure.
Jeremy Smith, Governor’s Chair at the University of Tennessee and director of the UT/ORNL Center for Molecular Biophysics, as well as experts from IBM and NVIDIA (companies that provide underlying technological components that help power Summit) recently briefed Nextgov on the work.
“Summit is the world’s most powerful supercomputer and was needed to rapidly get the simulation results we were looking for,” Smith said.
Related: Pentagon Suspends Troop Travel in Affected Countries; Trump Limits Travel from Mainland Europe
Energy in 2014 partnered with IBM, NVIDIA and Mellanox in a contract worth hundreds of millions to produce the Summit supercomputer to enhance civilian-focused scientific research, and also the Sierra supercomputer, which was designed for nuclear weapons simulations and is located at Lawrence Livermore National Laboratory. An Oak Ridge official confirmed Wednesday that Summit had not previously been used to address public health emergencies or virus outbreaks in the past.
But following the speedy outbreak of the new coronavirus strain, Smith found a relevant study recently published by Chinese scientists and realized he and his team “could take their work further,” with Summit’s help.
“That was on January 30th,” Smith said. “Micholas Smith, the other co-author on this paper, [in which the two Smiths published their initial work] promptly fell ill with the flu, but while he was incapacitated, still managed to run jobs on Summit.”
To infect cells, viruses bind to what’s called a “spike” protein and inject their genetic material into host cells. However, if particular drug compounds bind to the virus’ spike proteins, they could potentially block COVID-19 from infecting humans. The challenge is that researchers must embark on a very slow, intensive process to narrow down the range of potential variables that could work in blocking the virus, and each variable could hold millions to billions of unique pieces of data. But during a global health outbreak, speed is of the essence.
“In any situation where time is a critical factor—for example, a health issue or a company trying to get a new product out to market—supercomputers can play a critical role through digital simulation. It would be prohibitively time-consuming for researchers to grow virus cultures and test them with the near infinite amount of molecular compounds that exist in the physical world,” IBM’s Vice President, Exascale Systems Dave Turek told Nextgov. “So supercomputers like Summit can speed up that process by digitally simulating the effect to narrow down the range of possibilities researchers can look into.”
To put the “world’s smartest” supercomputer’s capabilities into perspective, Turek noted that “if every person on Earth completed one calculation per second, it would take 305 days to do what Summit can do in 1 second!”
Harnessing that power, Smith and his team designed a computational model then performed simulations on—and were subsequently able to rank—a database of 8,000 existing drug compounds that “are likely to be rapidly deployable because their toxicity profiles are mostly known.” Through that research and with the supercomputer’s help, the team was then able to narrow down the compounds, which include medications and natural compounds, to a “top-ranked” subset of at least 77 that might impair the new coronavirus’ strain from infecting host cells. The team can now run more targeted experiments on those, to advance and accelerate their efforts to find new solutions to cure and control COVID-19.
Without Summit’s powerful supercomputing and simulation capabilities, Smith said the work “would have been possible—but would have taken months rather than days.”
Experimentation generally follows computation, so now, Smith and his team must collaborate with relevant experimentalists to test those top compounds on the virus to see what works and what doesn’t, and ideally from there, eventually identify a drug that can be used to fight it. He said it is an iterative process, through which the team’s calculations are refined continuously to improve the predictions about how drug compounds will interact with the virus.
“Our hope is that, by using a database of known compounds, we can greatly reduce the time it takes to make an effective drug publicly available, but there is no guarantee,” Smith said.
Summit’s massive data processing capability was made possible in part by server nodes equipped with central processing units, or CPUs, from IBM and graphics processing units, or GPUs, from NVIDIA. Paresh Kharya, NVIDIA’s director of product management for artificial intelligence and cloud computing, told Nextgovthat the company’s GPUs have also been “instrumental” in helping other researchers understand the physical structure of the COVID-19 virus.
“Researchers at the University of Texas at Austin and the National Institutes of Health announced that they created the first 3D, atomic scale map of the part of the virus that attaches to and infects human cells—known as the spike protein,” Kharya said. Smith and his team now also plan to run their computational study again with this new, “highly accurate” version of the spike protein.
“The paper we submitted is a first step in a back-and-forth between experiment and computation that hopefully will find a drug to repurpose,” Smith said. “That could happen next week, next year, or never. If not, we’ll have to design a new one from scratch and that takes a long time and a lot of money because of the need for extensive safety testing.”
IBM’s Turek emphasized that a sort of all-hands-on-deck approach made possible through public-private partnerships “like the one [IBM has] with Oak Ridge, NVIDIA, and Mellanox” are paramount in paving the way for new scientific breakthrough, because they open many relevant experts up to a “wealth of new technology, perspective, insights, and ideas that we may never realize if we operated in a silo.”
“It would be foolish for any one organization to claim a monopoly on innovation or say that they have all of the ideas for the next breakthrough, IBM included,” Turek said. “While this project involves molecular biology, other research done on Summit includes how humans can survive a mission to Mars, understanding the genetic markers of substance addiction, and finding the origins of the universe. All of this wouldn’t be possible without the diverse minds that can access Summit via this public-private partnership.”
Smith, too, added that on top of this work shared between government and industry, “strong partnerships among the national laboratories and universities are important to the scientific infrastructure, enabling universities access to the world’s most powerful facilities and linking national laboratory experts with university researchers to help further their goals of scientific discovery.”
He noted that while there hasn’t been any discoveries from this initial research yet, the biggest impact so far is that the team has started that crucial interplay between computation and experiment.
“By the way, the calculations are still running as we speak, as we try constantly to improve the predictions,” Smith said. “We’ll keep plugging away.”