Folding@home’s crowdsourced network of volunteer computers has boomed during the pandemic, now comprising some one million citizen scientists who regularly donate their idle computing time to folding the proteins of SARS-CoV-2 in support of critical research. Folding@home was one of the earliest (and loudest) movers in large-scale computing when COVID-19 began picking up speed: as early as February, Folding@home was issuing a call to action and beginning simulations of coronavirus proteins. Eight months later, Folding@home’s detailed protein modeling has contributed to a wide range of research activities. In a recent blog post, Nvidia highlighted some of the important research that has emerged from Folding@home’s ongoing work on the coronavirus.
Folding@home simulated SARS-CoV-2’s proteins for a tenth of a second. While that may sound like a short time, it was a massive accomplishment, incorporating more than 442,000 atoms in constant motion for a single protein – enough to generate hundreds of billions of data points within a single microsecond. Folding@home calls the resulting simulations the largest collection of molecular simulations in history.
The work was groundbreaking, allowing for researchers to accurately characterize some of SARS-CoV-2’s specific functionality – such as the motion of its critical spike protein – and identify a wide range of “pockets” that the researchers say expand options for the design of drugs to fight COVID-19.
“We’ve simulated nearly the entire proteome of the virus and discovered more than 50 new and novel targets to aid in the design of antivirals,” said Max Zimmerman, a postdoctoral fellow at the Washington University School of Medicine in St. Louis (which hosts one of the labs supporting Folding@home). “We have also been simulating drug candidates in known targets, screening over 50,000 compounds to identify 300 drug candidates.”
This research – which was just released as a preprint last week – produced a gargantuan dataset comprising those hundreds of billions of atomic movements for that critical tenth of a second.
Where did that dataset end up? The answer, in part: the desk of Peter Messmer, leader of a scientific visualization team at Nvidia, who says he was approached to do something with the data “using more than the typical scientific tools to make it really shine.”
Messmer processed the data using Visual Molecular Dynamics and Nvidia Omniverse, a yet-to-be-released tool for collaborative 3D graphics and simulation. He then linked Maya (an animation suite typically used for movies and video games) to Omniverse in order to “fly” a camera through the proteins, which had been given surface properties to enable better visual recognition.
“Researchers are not confined to scientific visualization tools [with Omniverse],” Messmer said. “They can use the same tools the best artists and movie makers use to deliver a cinematic rendering – we’re bringing these two worlds together.”
The result: a haunting, otherworldly Magic School Bus trip through the living canyons of a deadly virus.
“Nvidia GPUs have been instrumental in generating our datasets, and now those GPUs running Omniverse are helping us see our work in a new and vivid way,” Zimmerman said. “I’ve been repeatedly amazed with the unprecedented scale of scientific collaborations.”