During SC20 in November, the HPCwire Readers’ and Editors’ Choice awards program celebrated its 17th year of honoring outstanding achievements in high-performance computing. Unlike other years, however, SC was conducted virtually, owing to safety concerns from the COVID-19 pandemic.
HPCwire’s awards looked a lot different this year, too – not only in that they were presented online, but also through the incorporation of a new category: Best Use of HPC in Response to Societal Plights. The inaugural year of the new category honors the myriad ways in which HPC was used by doctors, scientists and researchers to fight back against the pandemic from the moment it appeared to the present-day challenges.
Beyond the new category, COVID-19-related research won five additional awards across four more categories, for a total of eight awards across five categories. The awardees – spanning industry, academia and government – have helped medical researchers, policy-makers and individuals understand the virus and take steps to prevent its spread, and are celebrated in this video and in the article below.
Providing the hardware
The pandemic resulted in an unprecedented marshalling of HPC resources in order to tackle the Herculean computational tasks that faced medical and epidemiological researchers throughout the year. One of the earliest movers on this front was Folding@home, a crowdsourced research computing service that aggregates computational power provided by citizen scientists. Folding@home achieved historic levels of aggregate power during the pandemic thanks to millions of volunteer CPUs and GPUs – and then used that power to simulate the movements of SARS-CoV-2’s critical proteins, a key task for drug design. For this work, Folding@home won the Readers’ Choice Award for Best Use of HPC in Response to Societal Plights.
“I’m delighted to represent the entire Folding@home community in accepting the … award,” said Greg Bowman, director of Folding@home. “In under three months we had over a million new devices join the Folding@home project[.] With this, we were able to understand these moving parts of every possible protein from the virus and capture large structural changes[.]”
The top supercomputer in the world, Fugaku, was also an early mover – but not quite in the same way. The behemoth machine, which again topped the latest Top500 list at 442 Linpack petaflops, wasn’t due to arrive for some time… but researchers at Riken in Japan managed to launch it a full year ahead of schedule in order to use its record-shattering power to conduct a wide range of COVID-19 research. Since then, Fugaku has been churning out results, including detailed simulations of how viral droplets spread through masks and face shields and within densely packed spaces. The early riser machine and its operators won one of two Editors’ Choice Awards for Best Use of HPC in Response to Societal Plights.
In the private sector, meanwhile, AMD and partners nabbed the remaining Editors’ Choice Award for Best Use of HPC in Response to Societal Plights for its COVID-19 HPC Fund. At awards time, the fund had contributed 12 petaflops of cloud and on-premises computing resources to more than 20 recipient organizations conducting COVID-19 research.
Perhaps the most massive supercomputing coordination effort came from the COVID-19 HPC Consortium. Led by the U.S. Department of Energy (DOE), IBM and the White House Office of Science and Technology Policy (OSTP), the sprawling collaboration organized dozens upon dozens of major partners and allocated their supercomputing resources to researchers tackling the pandemic. The COVID-19 HPC Consortium won the Editors’ Choice Award for Best HPC Collaboration.
“I found this to be a very rewarding experience,” Barb Helland, a program manager at the DOE, told HPCwire. “To be involved in this effort, and to watch OSTP, DOE labs, NSF and NASA centers come together very quickly with universities and U.S. industries … with little more than a handshake to use our capabilities to help understand and find solutions to respond to COVID-19.”
Understanding the virus – and the disease
Some researchers focused on understanding the virus in all its forms. In the UK, the National Health Service worked with other public health agencies, the Wellcome Sanger Institute and more than a dozen academic partners to form the COVID-19 Genomics UK (COG-UK) Consortium. The consortium, powered by hardware from DDN, Dell and Lenovo, conducted real-time sequencing of tens of thousands of samples from positive COVID-19 cases in the UK. The COG-UK Consortium was awarded the Readers’ Choice Award for Best HPC Collaboration.
Beyond understanding the physical structure of the virus, other organizations performed crucial research on how the disease presented itself in patients. At Oak Ridge National Laboratory, researchers compared genes of cells from COVID-19-infected patients with those of non-infected patients, uncovering signs of a COVID-19-inducted “bradykinin storm” – a phenomenon that could explain some of the more bizarre symptoms caused by the deadly virus. This important discovery won the Oak Ridge researchers the Editors’ Choice Award for Top HPC-Enabled Scientific Achievement. “We’re really glad that our model for COVID-19 pathogenesis is having such a large impact on the supercomputing and scientific communities,” Dan Jacobson, chief scientist for computational systems biology at ORNL, told HPCwire.
Hunting for treatments
With computing resources in-hand, one of the most prominent – and computationally daunting – tasks early in the pandemic involved screening billions upon billions of molecules to see how, if at all, they interacted with or inhibited the key proteins of SARS-CoV-2: predominantly, its notorious spike protein. At Harvard Medical School, researchers ran VirtualFlow on Google Cloud to conduct one of the largest in silico drug screenings in history. This research (which used some 75 million CPU hours) won those researchers the Readers’ Choice Award for Best Use of HPC in the Cloud.
Similarly, at Oak Ridge National Laboratory, researchers used the Summit supercomputer – the most powerful publicly ranked system in the U.S. – to conduct large-scale screenings of potential drug molecules. They identified 77 small molecules with potential for therapeutic use against COVID-19, earning them the Readers’ Choice Award for Best Use of HPC in Life Sciences.
Learn more – and what’s next
To learn more about the winners of this year’s HPCwire Readers’ and Editors’ Choice awards, check out the awards presentations and acceptance videos here – and don’t forget to watch HPCwire’s video highlighting the award-winning COVID-19 research discussed in this article.
Though the year is coming to a close, the pandemic is far from over, and HPCwire’s coverage of the computational fight against the virus is ongoing, with supercomputers playing crucial roles in stemming the massive spikes in COVID-19 infections and in distributing limited vaccine doses. Stay tuned to learn more about HPC’s ever-growing role in public health and drug development.