How GPUs Keep Pace with Shape-Shifting Viruses

by Tonie Hansen

Editor’s note: This is the last of a series of five posts profiling finalists for NVIDIA’s 2015 Global Impact Award, which provides $150,000 to researchers using NVIDIA technology for groundbreaking work that addresses social, humanitarian and environmental problems.

Viruses are a threat to human health on a global scale. They mutate as fast as the vaccines and drugs designed to combat them.

The ability of the human immunodeficiency virus, or HIV, to adapt to antiviral drugs and treatments means new compounds need to be developed continuously.

HIV Capsid
HIV capsid

To keep pace with this fast-changing virus and others like it, scientists at the University of Illinois at Urbana-Champaign looked beyond the microscope and turned to supercomputers to pull data from multiple sources for analysis.

Examining how the complex communication game between the cell and the virus works could lead to therapies that disrupt these interactions.

“We’re using computers at the same time as working on experiments – it’s a close collaboration between people taking data and using computational means,” Klaus Schulten, professor of physics at the University of Illinois at Urbana-Champaign. “Computers are also our research instruments.”

The research has placed the university among five finalists for NVIDIA’s 2015 Global Impact Award. We award our annual grant of $150,000 to researchers using NVIDIA technology for groundbreaking work addressing social, humanitarian and environmental problems.

Chemical Warfare

Scientists harnessing GPU-powered supercomputers combine new methodologies and algorithms for molecular simulation, visualization and analysis, which can simulate complex human viruses such as HIV in full atomic detail.

The large size of the HIV capsid, or shell, combined with its irregular shape poses huge challenges in resolving its chemical structure.

“The communication of the capsid’s chemistry and the chemistry of the infected cell is very complex,” Schulten said. “There are 60 million atoms of the physiological fluids around the capsid in addition to the capsid’s own 4 million atoms. Their simulation requires very powerful algorithms. The only way we could do this was with supercomputers.”

For HIV to infect a cell, the virus must enter the cell and communicate chemically while evading attack by the cell’s defense mechanisms. These defense systems want to break the HIV capsid apart before it causes harm and inserts the viral genetic message into the cell’s genome.

“There’s chemical warfare going on, and that’s what we need to understand to develop new drugs,” Schulten said.

“Simulation of viral infections is very health-relevant if you think of flu season, the measles outbreak and Ebola,” he said. “We have to find vaccinations and treatments, and for that to happen, we need to understand how things get infected because a virus is very smart.”

Four capsids. The large size of the capsid, or shell, combined with its irregular shape poses huge challenges in resolving its chemical structure.
Four capsids. The large size of the capsid, or shell, combined with its irregular shape poses huge challenges in resolving its chemical structure.

Communication Game

For assessing the dynamics of the HIV capsid in the cell, scientists apply massive computing power that allows them to examine the chemistry of cell-virus capsid communication occurring during the infection cycle.

Work by the researchers led to the use of GPUs in the fastest computers in the world, including Titan, at Oak Ridge National Laboratory, and Blue Waters, at the National Center for Supercomputing Applications at the University of Illinois.

“We needed our computation accelerated by GPUs, which made them 3X faster,” Schulten said. “We’re intensely working on employing also the next generation of GPUs so we can apply our simulations to further types of viruses infecting the cell.”

Processing data for the complex structures of cells would have taken a powerful single-processor computer almost two decades, and “now it takes 20 minutes,” said Schulten.

Schulten and his team are now preparing for the next steps in technology, with computers that will become available in 2018.