It may be 20 years old, but Folding@home has never been more relevant. It serves as a shining example of the possibility of using massive amounts of data, unprecedented levels of computing power and a diverse community of researchers and citizen-scientists to find new ways to cure diseases.
Folding@home is a distributed supercomputer that harnesses the power of devices around the world to run simulations of protein dynamics, gathering clues about diseases like COVID-19, Ebola, Alzheimer’s, cancer, and more. The program has grown to include 2.5 million devices, making it the first supercomputer to reach 1 exaFLOPS of computing power.
This crowdsourced model is the key to its power. Anyone with a computer and an internet connection can download the Folding@home software and volunteer part of their computing power. When their computer is idle, the program runs simulations in the background and sends the results back to a dedicated server, where researchers comb through the results.
“We share the data we generate with anyone who wants it,” says Greg Bowman, director of Folding@home and associate professor of biochemistry and molecular biophysics at Washington University School of Medicine in St. Louis. “Certainly the scientific community is interested in mining the data, but we want to enable citizen-scientists to analyze it, too.”
It’s this ability to bring together a “diversity of ideas” that’s most exciting to Philip R.O. Payne, PhD, associate dean for Health Information and Data Science, chief data scientist, and director of the Institute for Informatics (I2). “There are so many complex problems that a single person or even institution can’t solve alone — these are the types of challenges that require us to pursue research as a team sport,” he says. “Who knows if somebody in high school or at a tech startup might come up with a novel solution to these types of challenges and how that could change our understanding of a disease. Folding@home allows us to create this economy of diverse ideas.”
Zeroing in on Proteins at the Atomic Level
The questions that are being asked through Folding@home and the data that comes out of them revolve around proteins — what they do, how they work, and, in particular, how they fold, which is to say how they assemble themselves before they do their job. That’s because proteins carry out all of the processes that make biology work, and if they make a mistake, or misfold, the result can lead to disease, like Alzheimer’s, Parkinson’s, Huntington’s, and many cancers.
Proteins also play a role in COVID-19, which is why Folding@home has pivoted its enormous supercomputing power to focus on 14 different proteins from the SARS-CoV-2 virus. One in particular is the COVID-19 spike, which Bowman and his team have nicknamed the COVID-19 Demogorgon because of the way it looks when it opens up its three receptor-binding domains to bind to ACE2, the protein found on the surface of human cells. Folding@home is running simulations of the COVID-19 Demogorgon opening its mouth to find potential therapeutic options, such as drugs that could block it from binding with human cells, preventing infection.
Folding@home makes it possible to test tens of thousands of potential drug designs nearly simultaneously. The candidates are then ranked on how effective they’re bound to be based on the simulations completed. Researchers can then take the most promising contenders out of the virtual world and into the real-world via laboratory testing.
“The cost to do this kind of testing can quickly get beyond what an academic group is able to do,” Bowman says. “This is a powerful way of focusing physical resources on the things that are most likely to be fruitful.”
The Broader Implications for Healthcare
Before the pandemic hit, Folding@home was tackling a range of health concerns across a broad spectrum of diseases. One of the recent findings produced through use of the platform has the potential to help in the fight against antibiotic resistance. Its simulations found a previously unknown pocket within β-lactamase, a protein that plays a key role in antibiotic resistant infections. This knowledge is helping researchers design new drugs to thwart it. Folding@home researchers have also used the supercomputer’s powers to investigate why certain mutations increase a person’s risk for Alzheimer’s disease and how those answers could help inform new strategies for therapeutics.
“By running these simulations, we can find unexpected binding sites for drugs that people hadn’t thought to go looking for before,” Bowman says. “With Ebola, we were able to look at a protein structure where we didn’t think there’d be a place to bind and find that it actually has a pocket that forms.”
Payne agrees that Folding@home’s capabilities as a high-throughput hypothesis generator is a major source of its value. “We’re drowning in a sea of data from bio-molecular instruments, electronic health records, and wearables, to name a few of the many examples, and yet we tend to ask and answer a very small number of questions using that data,” Payne says. “If we’re going to realize the benefits of all of these sources of data, we need to ask all of the relevant questions we can. The next 10 years of informatics and data science must focus on how we do something meaningful with such big data, and Folding@home is a foundation for that type of work.”
Help ensure that this important work continues by making a gift.