Inside the Global Race to Fight COVID-19 Using the World’s Fastest Supercomputers (via Qpute.com)
As the director of a global research organization, I feel obligated to use all the resources of cutting-edge science and technology at our disposal to fight this scourge. As a father, I want a lasting solution, one that serves not just in this crisis, but the next. And, as an American and a Spaniard, with family in two hot spots, I want to help. It’s as simple as that.
It started with a phone call to the White House on Tuesday, March 17, one that proved to be a catalytic moment for industry, academia and government to act together. This was the same week I received news from my mother that my cousin in Spain had tested positive for coronavirus. She’s a doctor and, just like all medical staff around the world right now, is on the front lines of the fight against this disease. This fight is personal for so many of us.
COVID-19 is deadly serious. This respiratory disease is triggered by a virus from the family of coronaviruses, which was identified in the 1960s but had never made such an assault on humanity. The virus prevents its victims from breathing normally, making them gasp for air. Fever, cough, a sore throat and a feeling of overwhelming fatigue and helplessness follow. Lucky ones recover within a few days; some show only mild or moderately severe symptoms. But some patients are not that lucky. Bulldozing its way through the body, the virus makes the lungs fill up with fluid, and may lead to a rapid death. No one is immune. While the elderly and those with underlying health conditions are more at risk, COVID-19 has taken the lives of people of all ages, some in seemingly good health. The disease is bringing our world to its knees.
But we are resilient, and we are fighting back with all the tools we have, including some of the most sophisticated supercomputers we have ever built. These machines—more than 25 U.S.-based supercomputers with more than 400 petaflops of computing power—are now available for free to scientists searching for a vaccine or treatment against the virus, through the COVID-19 High Performance Computing Consortium.
It was created with government, academia and industry—including competitors, working side by side. IBM is co-leading the effort with the U.S. Department of Energy, which operates the National Laboratories of the United States. Google, Microsoft, Amazon and Hewlett Packard Enterprise have joined, as well as NASA, the National Science Foundation, Pittsburgh Supercomputing Center and six National Labs—Lawrence Livermore, Lawrence Berkeley, Argonne, Los Alamos, Oak Ridge and Sandia, and others. And then there are academic institutions, including MIT; Rensselaer Polytechnic Institute; the University of Texas, Austin; and the University of California, San Diego.
The supercomputers will run a myriad of calculations in epidemiology, bioinformatics and molecular modeling, in a bid to drastically cut the time of discovery of new molecules that could lead to a vaccine. Having received proposals from all over the world, we have already reviewed, approved and matched 15 projects to the right supercomputers. More will follow.
But just a few days ago none of this existed.
On March 17, I called Michael Kratsios, the U.S. government’s chief technology officer. Embracing the potential of a supercomputing consortium, he immediately started mobilizing his team, including Jake Taylor, assistant director for quantum information science at the White House Office of Science and Technology Policy. Jake called major U.S. players that have high-performance computers and invited them on board. From the IBM side, Mike Rosenfield, whose team has designed and built multiple generations of world-leading supercomputers, partnered with RPI, MIT and the key computing leaders of the U.S. National Laboratories. The U.S. Department of Energy has been a partner from the very beginning, at the heart of it all.
Within 24 hours of that first call, collaborators outlined what it meant to be involved. We brainstormed how we would communicate to research labs worldwide what we could offer in terms of hardware, software and human experts, and how we would get them to submit proposals, and get those matched with just the right supercomputer.
Forty-eight hours passed. On Thursday, March 19, we set up the scientific review committee and the computing matching committee to manage proposals. At least one person from each of the members of the consortium had to be part of the process, all acting fairly and equally. From IBM, Ajay Royyuru joined the merit review committee; he is the leader of our Healthcare and Life Sciences research and together with his team has long been developing novel technologies to fight cancer and infectious diseases.
Ajay, too, has a personal stake in fighting back against COVID-19. In January, his elderly father passed away following a pulmonary illness. Ajay shares his house with his 82-year-old mother, and he worries about keeping her safe from this risk, just like so many of us worry about our parents. His extended family in India is now also confronting the unfolding of the pandemic.
On March 22, less than a week after the first discussion with Kratsios, the White House announced the consortium. Everyone knew that the clock was ticking.
It is still very early days, but Ajay and other reviewers can clearly see from the first wave of proposals that scientists are trying to attack the virus on all fronts—from drug discovery and development with AI-led simulations to genomics, epidemiology and health systems response. We need to understand the whole life cycle of this virus, all the gearboxes that drive it—how it encounters and infects the host cell and replicates inside it, preventing it from producing vital particles. We need to know the molecular components, the proteins involved in the virus’ biochemistry—then to use computational modeling to see how we can interrupt the cycle. That’s the standard scientific methodology of drug discovery, but we want to amplify it.
The virus has been exploding in humans for months now, providing an abundance of samples for computer modeling and analysis. Scientists are already depositing them into public data sources such as GenBank and Protein Data Bank. There are many unknowns and assumptions—but, Ajay tells me, a lot of proposals involve using the available protein structures to try and come up with potential molecular compounds that could lead to a therapeutic or a vaccine.
That’s already happening. Even before we formed the consortium, researchers at Oak Ridge National Laboratory and the University of Tennessee simulated 8,000 compounds and found 77 molecules that could potentially disarm the virus. But 77 is still a big number and running tests to find the correct molecule may take months. Here, my colleague Alessandro Curioni, an Italian chemist who heads IBM Research Europe and who had to self-isolate due to possible exposure to COVID-19, had an idea on how to speed things up.
In a conversation with European Commission executives in early March, Alessandro learned about an Italian pharmaceutical company, Dompé Farmaceutici and the E.U.-financed project they were working on. Last week, he orchestrated a meeting between its scientists and Oak Ridge, suggesting to both parties that they submit a joint proposal to the consortium. Perhaps together, with the help of supercomputers, they can reduce the number of the promising compounds from 77 to 10, five and, finally, one.
Humanity has more tools at its disposal in this pandemic than ever before. With data, supercomputers and artificial intelligence, and in the future, quantum computing, we will create an era of accelerated discovery. The consortium is an example of a unique partnership approach, and it shows that the bigger the challenge, the more we need each other.
Privacy & Cookies Policy
Necessary cookies are absolutely essential for the website to function properly. This category only includes cookies that ensures basic functionalities and security features of the website. These cookies do not store any personal information.
Any cookies that may not be particularly necessary for the website to function and is used specifically to collect user personal data via analytics, ads, other embedded contents are termed as non-necessary cookies. It is mandatory to procure user consent prior to running these cookies on your website.