As COVID-19 continues its devastating global spread, the scientific computing community is fighting back by bringing its considerable forces to bear against the virus. Joining the growing list of efforts responding to the COVID-19 pandemic, a team from the University of Alabama in Huntsville (UAH), led by Dr. Jerome Baudry, is using HPE’s Sentinel supercomputer via the Microsoft Azure cloud to identify potential COVID-19 drug treatments.
The work is described in a blog post from Joseph George, executive director of HPC Strategic Alliances, at Hewlett Packard Enterprise (HPE).
Dr. Jerome Baudry (credit: Michael Mercier | UAH)
“Dr. Baudry, a distinguished researcher in molecular biophysics, is taking a molecular docking approach to predict the binding – or ‘lock and key’ interaction – between the viral protein in the coronavirus and molecules that can counterattack its functions,” writes George.
The computational capacity provided by the Cray XC50 Intel-based “Sentinel” system, located within Azure’s Texas datacenter, is accelerating Baudry’s research, reducing the time it takes to evaluate candidate molecules for anti-COVID effects.
“At HPE, we believe in being a force for good, so when the COVID-19 pandemic struck, HPE quickly made available supercomputing resources, along with a dedicated technical staff, free of charge to help scientists tackle complex research. That’s when we met Dr. Baudry and set him and his team up on HPE’s Sentinel supercomputer, which can perform 147 trillion floating point operations per second and store 830 terabytes of data,” said George.
The HPE partnership supports Boundry’s work with the National Center for Natural Products Research (NCNPR) at the University of Mississippi School of Pharmacy.
Baudry and his team investigated 20,000 diverse molecules out of a larger set of 200,000 natural compounds, assessing the ability of the molecules to counter the functions of the COVID-19 protein. Leveraging Sentinel’s 1,920 cores, they modeled the docking of these target molecules with the virus’ protein, using 10 different orientations per molecule. A total of 200,000 molecular docking calculations was conducted in 10 hours, condensing what would previously have been a multi-week project down to one day.
“We use the laws of physics and chemistry with the Sentinel supercomputer to predict if a given natural product can ‘block’ the proteins that the virus needs to infect the cells and to replicate,” Dr. Baudry said.
“If we can identify computationally such a natural product,” he said, “then we have colleagues who will test it in specialized labs.”
Sentinel was introduced by Cray (now part of HPE) in 2018 as part of an “early access program” to support the onboarding of large-scale simulation and modeling workloads in the cloud with access to Azure services. The announcement followed Cray’s partnering with Azure in 2017 to offer Cray supercomputers and Cray Clusterstor storage systems in the Azure cloud, enabling customers to run their HPC and AI applications alongside their other cloud workloads.
A supporter of the U.S. COVID-19 HPC Consortium, HPE is involved in a number of initiatives that engage the company’s HPC systems and experience, detailed in this post from Peter Ungaro, senior VP and general manager of the HPC and Mission Critical Solutions business unit at HPE.