Alongside the White House Office of Science and Technology Policy (OSTP), IBM announced in March that it might assist coordinate an effort to offer lots of of petaflops of compute to scientists researching the coronavirus. As a part of the newly launched COVID-19 High Performance Computing (HPC) Consortium, IBM pledged to help in evaluating proposals and to offer entry to assets for tasks that “make the most immediate impact.”

Much work stays, however among the Consortium’s most distinguished members — amongst them Microsoft, Intel, and Nvidia — declare that progress is being made.

Petaflops of compute

Powerful computer systems permit researchers to undertake excessive volumes of calculations in epidemiology, bioinformatics, and molecular modeling, lots of which might take months on conventional computing platforms (or years if finished by hand). Moreover, as a result of the computer systems can be found within the cloud, they allow groups to collaborate from wherever on the planet.

Insights generated by the experiments may help advance our understanding of key points of COVID-19, equivalent to viral-human interplay, viral construction and performance, small molecule design, drug repurposing, and affected person trajectory and outcomes. “Technology is a critical part of the COVID-19 research going on right now all over the world,” Thierry Pellegrino, VP of Dell Technologies, a member of the Consortium, informed VentureBeat. “It’s crucial to the population of our planet that researchers have the tools to understand, treat, and fight this virus. Researchers around the world are true heroes doing important work under extreme and unfamiliar circumstances, and we couldn’t be prouder to support their efforts.”

VB Transform 2020 Online – July 15-17. Join main AI executives: Register for the free livestream.

Companies and establishments matched 62 projects within the U.S., Germany, India, South Africa, Saudi Arabia, Croatia, Spain, the U.Okay., and different international locations with supercomputers from Google Cloud, Amazon Web Services (AWS), Microsoft Azure, IBM, and dozens of educational and nonprofit analysis establishments without cost. These are operating on over 136,000 nodes containing 5 million processor cores and greater than 50,000 graphics playing cards, which collectively ship over 483 petaflops (430 trillion floating-point operations per second) of compute throughout {hardware} maintained by the Consortium’s 40 companions.


In addition to supercomputing infrastructure constructed atop its Azure cloud computing platform, Microsoft is offering researchers with networking and storage assets built-in with workload orchestration by way of Azure HPC. Concurrent with that is the corporate’s AI for Health program, which in April allotted $20 million to developments in 5 key areas — information and insights, remedy and diagnostics, allocation of assets, dissemination of correct data, and scientific analysis — with the aim of bolstering work associated to COVID-19.

As part of its work with the Consortium, Microsoft says it’s offering groups entry to its scientists spanning AI, HPC, quantum computing, and different areas of pc science at Microsoft Research and elsewhere. Much of those researchers’ work to this point has entailed fundamental scientific discovery about COVID-19 itself and the way it interacts with the human host, together with the design of therapeutics, by way of:

  • Research simulations.
  • Molecular dynamics modeling.
  • 3D mapping of virus protein buildings.
  • Compound screening to see if current drug molecules are capable of inhibit mobile entry of the virus.

Microsoft says every group it collaborates with receives a full Azure HPC setting, together with Azure CycleCloud with the Slurm workload supervisor, best-fit Azure Virtual Machines, and storage. These are configured to scale on-demand and meet compute as crucial, they usually’re tailor-made to the particular analysis wants of the grantee.

Nepali modeling and ventilator splitting

Through the Consortium, Microsoft’s AI for Health is supporting the nonprofit analysis institute Nepal Applied Mathematics and Informatics Institute for Research (NAAMII), which is using simulation to mannequin how COVID-19 would unfold among the many Nepali inhabitants, given totally different situations. These fashions, Microsoft says, can present patterns that may doubtlessly save lives and livelihoods.

Duke University, one other grantee, is leveraging Azure to analyze ventilator splitting, a way that permits a number of sufferers to make use of the identical ventilator. The Matlab division of MathWorks teamed up with Microsoft to optimize the researchers’ evaluation for distributed computing environments.


Google continues to offer compute, storage, and workload administration companies to Consortium grantees by way of Google Cloud Platform, and it not too long ago put aside $20 million in computing credit for educational establishments and researchers finding out COVID-19 remedies, therapies, and vaccines. As part of its work with the Consortium, the corporate is collaborating on epidemiological modeling with Northeastern researchers and making use of AI to medical imaging with the Complutense University of Madrid.

Google additionally partnered with the Harvard Global Health Institute to fund firms, authorities businesses, nonprofit organizations, and establishments engaged on COVID-19 analysis. The tech large — together with Microsoft — additionally kicked off a program with Microsoft-backed cloud firm Rescale to supply HPC assets for free of charge to groups working to develop COVID-19 testing and vaccines. Rescale supplies the platform that researchers launch experiments and report outcomes on, whereas Google and Microsoft provide the backend computing assets.


Amazon, like Google, is supplying compute and instruments to researchers matched by way of the Consortium. Over 11 groups are at the moment utilizing its infrastructure, and devoted Amazon Web Services answer architects convention with the scientists each week.

As part of its AWS Diagnostic Development Initiative, Amazon can also be offering $20 million in computing credit to over 35 establishments and personal firms which might be leveraging AWS to additional the event of COVID-19 point-of-care diagnostics — i.e., testing that may be finished at house or at a clinic with same-day outcomes. “This is a global health emergency that will only be resolved by governments, businesses, academia, and individuals working together to better understand this virus and ultimately find a cure,” mentioned Teresa Carlson, VP of worldwide public sector at AWS, in a press release.

Developing protein decoys

At the MIT Media Lab, impressed by a researcher at Johns Hopkins University, a staff is figuring out “decoy” proteins of ACE2 receptors (the receptors coronaviruses bind to contained in the human physique) that may render COVID-19 inert. Using a machine studying mannequin educated on information concerning the ACE2 receptor and operating on AWS, the researchers try to foretell which variants of the decoy received’t work together with different proteins within the physique and trigger dangerous unwanted effects. If all goes effectively, checks in mice will begin quickly, with medical trials starting towards the tip of summer season.

In separate efforts, AWS is empowering researchers on the Children’s National Hospital to mix lots of of information units to determine genes that could be focused to deal with COVID-19. A staff at Iowa State University is tapping evolutionary fashions with public genomic information units to suss out the relationships between strains of COVID-19 to grasp how they mutate and unfold. And scientists at Emory University are growing a web-based software — tmCOVID — to extract and summarize key ideas in scientific research on COVID-19.


Nvidia says that 14 of the Consortium’s tasks have consumed over Three million GPU hours on the Nvidia-powered Summit supercomputer at Oak Ridge National Laboratory. Summit is the world’s quickest supercomputer, as ranked by the Top 500 list of supercomputers. And it additionally provides its personal 20,000-GPU infrastructure — SaturnV — which the corporate’s researchers are primarily utilizing to optimize COVID-19 analysis purposes

Nvidia has been utilizing extra cycles on SaturnV to run [email protected], a distributed computing mission that simulates protein dynamics in an effort to assist develop therapeutics for numerous ailments, together with COVID-19. It has assisted in matching researchers to supercomputers based mostly on every researcher’s particular necessities.

Quantum chemistry and digital screening

In partnership with Microsoft, Nvidia is working with the University of California, Riverside on quantum chemistry options that profit from GPU optimization. The variety of attainable COVID-19 inhibitors are immense, and finishing up experimental research on all of the candidates is each infeasible and cost-inefficient. The hope is that the mission’s predictive, GPU-enabled simulations — which use as much as 800,000 GPU hours on Azure — will present steerage for efforts that slim in on probably the most promising candidates.

According to Nvidia, in lower than every week its specialists helped mission lead Bryan Wong’s package deal analysis code utilizing HPC Container Maker, the corporate’s open supply software that ships with 30 containerized HPC purposes. And they tapped Nvidia’s Nsight debugging software to develop a repair for an onerous bug — making it attainable to perform work scheduled to take 800,000 GPU-hours in 300,000 GPU-hours for a financial savings of $500,000.

At Carnegie Mellon University, a staff led by Olexandr Isayev labored with Nvidia to use AI approaches to the duty of high-throughput digital screening, which makes use of algorithms to determine bioactive molecules. Unlike conventional scientific simulations, which take a brute drive strategy to issues by trying to simulate each attainable mixture of molecular interplay, AI makes educated guesses that cut back the variety of combos to be simulated. This results in theoretically quicker candidate drug discovery (and faster discipline trials). Isayev estimates that it could be as a lot as 1,000,000 instances quicker than typical mechanical calculations.

The first step within the course of is utilizing AI to investigate a library of molecules that may be bought from chemical firms, making ready them for screening in simulation. The greatest candidates from the screening will then be simulated utilizing AI-enhanced molecular dynamics, and high hits from the ultimate screening might be examined in accomplice laboratories.

At the conclusion fo their work, Isayev and colleagues plan to deposit their information units within the open source COVID-19 data lake, a centralized repository of curated information units maintained by Amazon’s AWS division in hopes that different researchers will profit from them.


IBM VP of technical computing Dave Turek says COVID-19 analysis continues with companions throughout the spectrum — on machines powered by its {hardware} and inside laboratories and establishments it has relationships with. “Without any large contracts or anything of the kind, [the Consortium] came together in a way to both share resources and manage a process of expediting the scientific proposals that came into the consortia and match it to the best resources,” he mentioned in a press release. “The teams are making rapid progress, and these supercomputing-powered projects are using novel approaches to understanding the virus.”

For instance, IBM researchers on the Hartree Centre in Daresbury, England partnered with University of Oxford scientists to mix molecular simulations with AI in discovering compounds that could possibly be repurposed as anti-COVID-19 medicine. Using Summit and the Texas Advanced Computing Center’s (TACC) Frontera, the fifth-fastest system per the Top 500, the staff says they’re conducting months of analysis in a matter of hours.

Generating molecular compounds

With the assistance of IBM, researchers on the University of Utah tapped the National Center for Supercomputing’s Blue Waters and TACC’s Longhorn and Frontera to generate greater than 2,000 molecular fashions of compounds related for COVID-19. They ranked the fashions based mostly on the molecules’ drive discipline power estimates, which they theorized may assist scientists design higher peptide inhibitors of an enzyme to cease COVID-19.

The staff investigated the construction of the virus’ essential protease, an enzyme that breaks down proteins and peptides, in advanced with a peptide inhibitor referred to as N3. They then utilized an strategy developed to determine Ebola-stopping molecules that includes molecular dynamic simulations and optimization of particular buildings. This enabled the COVID-19 protease to interrupt down a sequence of comparable, easy-to-detect probes that had already been designed, serving as the premise for assessments that check the inhibitors’ effectiveness.

The work constructed on a physique of information about how the potential power generated by atoms may give a molecule a positively or negatively charged “force field” that draws or repels different molecules. Using AMBER, a molecular dynamics code, the researchers noticed experimental outcomes inside one hundred-millionth of a centimeter, a measure imperceptible to all however the strongest microscopes.

The University of Utah’s Schmidt lab will later rework the peptide leads into biopharmaceutical scaffolds referred to as round modified peptides. “Our hope is that we find a new peptide inhibitor that can be experimentally verified in the next couple of weeks. And then we will engage in further design to make the peptide cyclic to make it more stable as a potential drug,” University of Utah professor and analysis lead Thomas Cheatham mentioned in a press release.

Mapping how COVID-19 spreads

It’s effectively understood that COVID-19 spreads by way of virus-laden droplets, that are transported round environments by air-con items, wind, and different types of turbulence. But airborne transmission charges stay a topic of rivalry, and some experts say gathering helpful proof of airborne transmission may take years and value many lives.

In a safer pursuit of readability, scientists at Utah State University, Lawrence Livermore National Lab, and the University of Illinois intend to make use of the Consortium’s supercomputing assets to review person-to-person transmission of airborne respiratory infectious ailments like COVID-19. They are working from the speculation that aerosolized droplets from human airways contaminate rooms extra shortly than initially assumed. They’ll leverage high-fidelity multiphase large-eddy simulations (LES) — mathematical fashions for turbulence utilized in computational fluid dynamics — operating on IBM {hardware} to find out cloud paths in typical hospital settings.

The short-term intention might be to grasp how lengthy a cloud persists and the place the particles settle, which may inform non-pharmacological methods to scale back the unfold. “The [goal] of this study is to fundamentally improve our understanding of person-to-person transmission of airborne respiratory infectious diseases,” wrote the researchers in a press release. “Our findings will [make] it safer for health care professionals.”

Studying genetic susceptibility

Beyond isolating COVID-19-killing compounds and mapping viral unfold of the virus, researchers try to outline threat teams by performing genome evaluation and IBM supercomputer-enhanced DNA sequencing.

A staff of scientists affiliated with NASA have noticed that COVID-19 seems to trigger pneumonia, triggering an inflammatory response within the lungs referred to as acute respiratory misery (ARDS). To check this, they plan to make use of the supercomputer at NASA’s Ames Research Center, which can sequence the genome on sufferers who develop ARDS and people who don’t. 

If all goes effectively, the staff believes their examine will end in sensible instruments for predicting which COVID-19 sufferers are more likely to develop ARDS, and due to this fact which sufferers are more likely to want intensive assist previous to the emergence of extreme signs. Such instruments may assist information intensive care useful resource utilization for the sickest sufferers, and allow well being care staff to raised handle ongoing remedy.


Intel is actively concerned within the design, improvement, and deployment of a number of Consortium-affiliated supercomputers, in addition to the upcoming Aurora at Argonne National Laboratory in Chicago. The firm says it has a employees of engineers engaged on code optimizations for HPC purposes that embody LAMMPS (a molecular dynamics code), Gromacs (a package deal for protein, lipids, and nucleic acids simulation), NAMD (one other molecular dynamics code), AMBER, and others. Intel can also be sharing instruments, structure information, and software program with companions to reinforce COVID-19 purposes and scale their efficiency on Intel-based {hardware}.

One particular space of focus for Intel is a collaboration with NAMD to launch a model of the code that gives quicker simulations on Xeon processors that assist AVX-512. The firm says the numerous efficiency enhance will permit researchers to realize longer timescales within the simulation of related molecules related to COVID-19, by extension enabling them to raised perceive points of viral an infection with “atom-level” element. The replace is predicted to be made public for early use in June.

Hewlett Packard Enterprise

Some of Hewlett Packard Enterprise’s (HPE) work is finished by way of the Consortium, whereas the remainder is targeted on numerous clients and companions. As a results of its acquisition of Cray in September 2019 for roughly $1.Three billion, HPE claims it now has extra supercomputers and HPC techniques in use than any main analysis heart.

“High-performance computing is more powerful today than it’s ever been, and its massive computing power — along with other advanced capabilities — has significantly transformed drug discovery,” mentioned Peter Ungaro, former Cray CEO and head of HPE’s HPC and mission-critical techniques group, in a press release. “Supercomputing and HPC systems unlock greater potential for AI and machine learning applications, and when applied to 3D modeling and simulations, dramatically [accelerate] time-to-insight and [increase] scientific outcomes. Our work within the consortium provides the researchers with HPC capabilities they wouldn’t normally have access to independently to help fast-track the discovery of a cure for the pandemic.”

Drug design analysis

In partnership with Microsoft, HPE is working with a staff on the University of Alabama in Huntsville (UAH) to provide its Sentinel supercomputer by way of the Azure cloud. With the supercomputer, together with a staff of devoted HPE specialists, it’s supporting numerous levels of the drug design course of at UAH.

The researchers are using a molecular docking strategy, a sort of bioinformatic modeling that includes the interplay of two or extra molecules to yield a steady mixture. Drawing on a big, open set of pure merchandise present in crops, animals, fungi, and the ocean, Sentinel is performing calculations to find out how pure compounds work together with COVID-19’s protein. Previously, 20,000 molecular dockings could possibly be improved in opposition to a protein goal in seven or eight minutes, versus the complete 24 hours it used to take. Now, the analysis staff can carry out as many as 1.2 million molecular dockings per day.

Elsewhere, HPE is supporting work on the Lawrence Livermore National Laboratory with the Theta supercomputer, which is housed on the Argonne Leadership Computing Facility. The researchers’ aim is to use AI to speed up the method of simulating billions of molecules from a database of drug candidates. They’ve narrowed down the variety of potential candidates from 1040 to a set of 20, they usually’ve tapped Catalyst — an HPE-powered HPC cluster that generates predictions like experimental and structural biology information  — to enhance outcomes and pace up discovery.

HPE can also be collaborating with France’s National Center for Scientific Research and GENCI to arm scientists on the Sorbonne University in Paris with GENCI’s Jean Zay supercomputer, which HPE designed. The staff is utilizing Jean Zay to optimize the Tinker-HP software program, an strategy to parallel computing enabled by a number of graphics playing cards and designed to simulate on the degree of atoms for big organic molecules. Tinker-HP concurrently performs a variety of data-intensive calculations to create 3D simulations of molecular interactions quicker and at excessive resolutions than would in any other case be attainable.

Contributions from the non-public sector

The nature of the Consortium’s work isn’t strictly educational. Startups hope to make use of the group’s huge computational assets to develop remedies, molecular designs, and medicines focusing on COVID-19.

Kolkata-based Novel Techsciences is figuring out phytochemicals from the greater than 3,000 medicinal plans and anti-viral plant extracts in India that may act as pure medicine in opposition to COVID-19. The staff additionally plans to isolate plant-derived compounds that would assist sort out multi-drug resistance that arises because the coronavirus mutates, with the aim of growing a complete prophylactic remedy regime.

In London, Y Combinator-backed PostEr is overseeing the Moonshot Project, which goals to supply inhibitors based mostly on over 60 fragment hits (i.e., molecules validated to bind to a goal protein, making them a chemical place to begin for drug discovery) which have been remoted in experiments to find out the molecular construction of COVID-19. By operating machine studying algorithms within the background to triage recommendations and generate synthesis plans, PostEra has recognized round 21 extremely efficient volunteer-submitted molecular designs, which might be synthesized by chemical firm Enamine. The outcomes of this mission might be examined on animals in months.

If profitable, PostEra’s could be one of many first medicine developed in an open supply trend. “[Machine learning] can reduce the time to determine optimal ways to make these compounds from weeks to days,” the corporate mentioned in a press release. “[We believe] the worldwide scientific community [can suggest] drug candidates that might bind to, and neutralize, [COVID-19].”

Another non-public sector mission is led by London-based AI startup Kuano. This staff’s intention is to realize insights from ailments which might be just like COVID-19 — primarily different coronaviruses — to design an efficient COVID-19 drug. This effort depends on a genetic algorithm that searches the chemical area surrounding current antiviral medicine and a deep learning-based classification mannequin constructed on current binding information. The firm is combining these instruments with docking and molecular dynamics simulations to reinforce the outcomes and yield machine studying fashions that can be utilized to attain molecular designs for synthesis as antiviral compounds.

As for AI and drug improvement startup Innoplexus, it’s additionally working with the Consortium’s supercomputers to speed up the invention of molecules that would result in a drug to fight COVID-19. It expects to run permutations on 5 promising candidates — particularly, candidates which might be potent, non-toxic, and will be manufactured.

Early days

Despite the truth that a lot of the work stays within the early levels, momentum across the Consortium seems to be accelerating.

Last month, IBM introduced that UK Research and Innovation (UKRI) and the Swiss National Supercomputer Center (CSCS) will be a part of the Consortium, making obtainable machines that embody the University of Edinburgh’s ARCHER; the Science and Technology Facilities Council’s DIRAC; the Biotechnology and Biological Sciences Research Council’s Earlham Institute; and Piz Daint, the sixth-ranked supercomputer on the planet, in accordance with the Top 500. The new additions have introduced the entire obtainable petaflops as much as 483 from 437 in May and 300 in mid-March.

“The COVID-19 HPC Consortium … is the largest public-private computing partnership ever created. What started as a series of phone calls […] five days later, more than two dozen partners came on board, many who are typically rivals,” mentioned IBM’s Turek. “Without any large contracts or anything of the kind, this group [is coming] together in a way to both share resources and manage a process of expediting the scientific proposals that came into the consortia and match it to the best resources.”