GPUs Help Dengue Fever Researchers Slow Time

by Karen Xia

Nicknamed “breakbone fever,” illnesses caused by the dengue virus are not only as painful as they sound, but also occur in hundreds of millions of people per year, killing about 20,000 annually.

By using NVIDIA Tesla GPUs, however, a research team at Colorado State University is one step closer to preventing the replication of the disease – which currently has no cure — in the human body.

For three decades, scientists have known that the dengue virus survives due to replication of an enzyme called NS3, found in the dengue virus. But until now, no one has been able to find “binding sites” where drugs can influence the enzyme and inhibit genome replication to prevent the virus from making copies.

Manipulating Time

Previous supercomputer simulations of NS3 weren’t much help to scientists. To find binding sites of the enzyme, researchers needed to visually simulate how NS3 works on a molecular scale. But the best timescales only ranged from tens to hundreds of nanoseconds – far too brief to precisely capture steps of the replication process.

These problems meant that drugs currently identified to inhibit NS3 weren’t specific enough to be truly effective, and could also have significant side effects due to interactions with other cellular proteins.

NVIDIA GPUs helped change what researchers could see.

In their study, researchers showed that San Diego Supercomputing Center’s Comet and Bridges supercomputers, equipped with Tesla GPUs, enabled the team to simulate enzyme motion more efficiently. In particular, advanced software and rapid calculation speeds slowed the simulations to microseconds — 100x longer than previously reported simulations — which would’ve been impossible with CPU-only supercomputers.

As a result, researchers found a set of amino acid residues in NS3 that relay messages between coenzymes and RNA like a tin can telephone — something no one had seen before.

Researchers say they’ll continue to use their model to look into NS3 enzyme behavior, which also plays a role in the Zika and West Nile viruses. Further studies, especially with increased precision and graphical capabilities, could one day prevent such diseases from spreading.

“The development of technology such as NVIDIA GPUs and the corresponding software, Amber, to utilize this hardware effectively has greatly enhanced our ability to do these simulations. Ideally we can continue to push hardware and software development to eventually probe millisecond timescales, which are even more relevant to biology,” said Martin McCullagh, an assistant professor at Colorado State University who worked on the study.