UW–Madison astrophysicists donate computing resources to aid COVID-19 research
The IceCube Neutrino Observatory is possibly the world’s strangest telescope. Located at the South Pole, it is made up of over 5,000 basketball-sized light sensors embedded in a cubic kilometer of ice. Thousands of computers back at the University of Wisconsin–Madison, IceCube’s lead institution, scour data from those sensors for evidence of elusive subatomic particles that originate in outer space: astrophysical neutrinos.
Now, some of these computing resources are being used to simulate something different—protein folding of SARS-CoV-2, the coronavirus responsible for COVID-19. How proteins fold into three-dimensional shapes is difficult to predict but has big effects on biological interactions, like those between a virus and its host. These simulations will help researchers understand how the virus compromises human immune systems and reproduces.
While IceCube remains operational, its home research center at UW–Madison, the Wisconsin IceCube Particle Astrophysics Center, WIPAC, is temporarily providing some of its available computing resources to Folding@home. This citizen-science distributed-computing project crowdsources computationally intensive tasks like simulating protein dynamics. Distributed computing projects like Folding@home combine the power of thousands of individual computers contributed by their owners to process different portions of data simultaneously, significantly speeding up their results.
“It just feels right to make the effort to share computing resources from fields as far removed from virology as neutrino astrophysics,” says Kael Hanson, director of WIPAC. “We’re pleased to aid in research that could ultimately lead everyone impacted by the current COVID-19 situation out of the crisis.”
Folding@home started in 1999, but it has recently seen a surge in interest as people seek ways to help researchers understand COVID-19. They include Benedikt Riedel, global computing coordinator for the IceCube Neutrino Observatory and computing manager at WIPAC.
Riedel had been in touch with his scientific computing collaborators since mid-March to discuss how they could help the COVID-19 effort. When he heard from them about Folding@home, Riedel suggested supporting it to WIPAC administration, who then received approval from IceCube’s primary funder, the National Science Foundation. Since the donated computing cycles primarily come from already available resources, they do not significantly hamper WIPAC or IceCube projects.
“These are unprecedented times, and I feel like we should do what we can to help other researchers,” says Riedel. “So far it is going well, and I am hoping that we can continue to donate even after this ends.”
IceCube’s computing resources consist of roughly 5,000 traditional computers using CPUs (central processing units) and 300 computers using GPUs (graphical processing units), each containing massive numbers of simple parallel computing elements. As a member of the Open Science Grid—a national distributed-computing partnership that provides high-throughput computing resources to science projects around the country—IceCube had already been sharing computing resources for other projects. Dozens of other Open Science Grid members are also contributing their resources to COVID-19 research.
WIPAC is able to contribute to Folding@home thanks to software that manages the allocation of the center’s diverse computing resources to different users’ competing computational tasks. Specifically, WIPAC uses tools from the HTCondor software suite that was developed and is maintained by UW–Madison’s Center for High Throughput Computing (CHTC) to effectively manage the computational workload.
“The long partnership between WIPAC and CHTC is founded on a commitment to share resources and knowledge,” says Miron Livny, director of the CHTC. “It is gratifying to see this partnership contributing to the computing challenge of protein folding of SARS-CoV-2.”
The first protein-folding simulations received high priority and will continue to be executed along with the day-in-day-out IceCube workload, which is continually submitted to the system. You can follow WIPAC’s contributions to Folding@home here.