Staying home isn’t the only way to help fight the coronavirus pandemic.
Hundreds of thousands of volunteers have added their home computers to a vast network that forms a virtual supercomputer called Folding@home. The Folding@home project, which uses crowdsourced computing power to run simulations of proteins for researchers studying diseases, announced in February that it would begin analyzing proteins found in the coronavirus behind the ongoing pandemic (SN: 3/4/20). These proteins are tools that help the virus infect human cells. Using computer simulations, researchers are mapping the coronavirus’s proteins, in hopes of revealing vulnerabilities that can be attacked with new drugs.
The more volunteers who donate their unused computing power to the effort, the faster the virtual supercomputer can work its magic. Since the project announced its new focus on the coronavirus, around 400,000 new volunteers had joined, as of March 19. By March 26, that number had swelled to around 700,000. The collective computing power of that legion of volunteers makes Folding@home by far the most powerful supercomputer in the world.
Science News spoke with project leader Gregory Bowman, a biophysicist at Washington University School of Medicine in St. Louis, about how the project works and how people can help.
How do simulations help map coronavirus proteins?
Researchers have taken snapshots of the proteins of the coronavirus, called SARS-CoV-2, using techniques like X-ray crystallography and cryo-electron microscopy (SN: 10/4/17). But proteins don’t hold still, Bowman says.
“All the atoms in the protein and [its surroundings] are continually pushing and pulling on each other,” he says. “What we’re doing is modeling those physical interactions in the computer.” Those simulations reveal the different shapes a protein’s structure can take.
What kinds of vulnerabilities are you looking for?
“You want a nice pocket on the surface of a protein where you can imagine this little molecule that we design inserting into a groove,” Bowman says. But many proteins, particularly those in viruses, have seemingly smooth surfaces, making them hard to target.
Folding@home simulations give scientists a chance to uncover what Bowman calls “cryptic pockets” — potential docking sites for drugs that aren’t visible in still images of the protein, but are revealed as the protein wriggles around in a computer simulation.
Has this worked for other viruses?
“We actually took one protein from the Ebola virus and ran simulations and discovered one of these cryptic pockets,” Bowman says. “Then we went and did the experiments to show that there really is a small pocket, and if we stick a small molecule in there, it really can shut the protein’s function off.” Likewise, a new drug molecule could be designed to stick in the chemical cogs of a SARS-CoV-2 protein that renders the virus unable to infect human cells.
Why not just find an existing drug that works for the coronavirus?
“That would be amazing,” Bowman says. Developing new drugs can take years or even decades, so researchers are investigating several existing drugs — such as those designed to fight hepatitis C, Ebola and malaria — as potential COVID-19 treatments (SN: 3/10/20). But “there’s no guarantees that these things will work,” he says. For instance, antiviral drugs used to treat HIV that initially looked promising showed no clear benefit for coronavirus patients in a recent clinical trial (SN: 3/19/20). Efforts like Folding@home supplement tests on existing drugs by expanding the search.
Even if someone does identify a drug that can cripple SARS-CoV-2, “we don’t want to stop there,” Bowman says. “The assumption is that, like many viruses, this is going to mutate pretty fast, and that if we don’t keep up with it, we’ll be right back with the same problem we have now. Tackling this thing on many fronts is our best bet for success.”
Why do you need a supercomputer for the simulations?
“We have to work on very, very, very small timescales” to capture the tiny jitters of atoms in proteins, Bowman says. “Each step in the simulation is on the order of a femtosecond,” or one quadrillionth of a second. To track protein motion over, say, a second, “we’ve got to do like a billion-squared operations on the computer, and each of those operations requires us to ask how every pair of atoms in the protein and surrounding solution are interacting with each other,” he says. By drawing on the computing power of many volunteers at once, Folding@home performs calculations in a single month that could take an ordinary desktop computer 100 years.
Folding@home isn’t the only supercomputer put to the task of studying SARS-CoV-2. On March 23, the White House announced a new consortium of companies, universities and government agencies — including several national laboratories, NASA, IBM and Microsoft — that are offering researchers access to their supercomputers to expedite the discovery of treatments or a vaccine for SARS-CoV-2.
Who can help with Folding@home?
“Anyone can install our software on their personal computers and contribute” some of their unused computing power, Bowman says. “We’ve got everyone from people running it on their older laptops, to gamers that have really hardcore machines to … businesses who are pointing computer clusters at Folding@home.”