Remdesivir is one of a handful of therapeutic antiviral drugs that have been proven to improve outcomes for COVID-19 patients, and as such, is a crucial weapon in the fight against the pandemic – especially in the absence of a vaccine. However, remdesivir is far from perfect: patients with severe cases of COVID-19 who are treated with remdesivir still show a high (though reduced) rate of death and serious complications. Now, a team of scientists at the University of North Texas are applying supercomputing power in an effort to improve remdesivir’s efficacy.
The researchers – led by G. Andrés Cisneros, a professor of chemistry – are investigating how remdesivir (and similar drugs) inhibit the “NSP-12” protein and the main protease in SARS-CoV-2, stunting its ability to copy itself. “We’re investigating how this process happens,” Cisneros said in an interview with Jorge Salazar of the Texas Advanced Computing Center (TACC). “By doing this, perhaps there might be a way for us and other scientists to come up with ideas on whether and how remdesivir can be improved.”
The research began when Cisneros saw newly released information about the structure of the polymerase containing NSP-12. “I contacted my group and told them that with this information, there’s something we can do to help with the pandemic,” he said. The team proposed the project to the COVID-19 HPC Consortium, which allocated the research time on two supercomputers at TACC.
The team is simulating these proteins – and the chemical reactions with them – using a combination of classical molecular dynamics and “QM/MM” (a hybrid approach combining quantum mechanics and molecular mechanics). “We’re talking about systems that we simulate that are in the hundreds of thousands of atoms,” Cisneros said. These calculations required immense supercomputer power, which the researchers found in the TACC systems: first, Stampede2 (10.7 Linpack petaflops), and later, Frontera (23.5 Linpack petaflops), which now hosts the bulk of the research.
“We were very fortunate to be granted an allocation on Frontera to be able to work on investigating the mechanism of drugs that target two specific proteins in COVID-19,” Cisneros said. “Frontera, with not only compute power but the intercommunication between the nodes, allows us to run these QM/MM calculations with much higher, not only speed, but also throughput.”
In terms of the team’s progress on the simulations: so far, so good. The team has reported that the models generated thus far are “very close” to the experimental structure of the proteins. “That’s really useful for us, because it validates the model that has been built by the group and shows that we are on the right track,” Cisernos said.
With that in hand, the researchers are investigating a series of six inhibitor molecules – and the first simulations are already in progress, with results expected in five to six months. “These are very expensive calculations,” Cisneros said. “Also, running the analysis takes time. If we were to use just the resources at home, it would take several years.”
To read the reporting from TACC’s Jorge Salazar, click here.