The University of Bristol in the UK, in collaboration with the Bansomdejchaopraya Rajabhat and Chulalongkorn Universities in Bangkok, used GPUs to help prevent the a possible outbreak of H1N1 which is immune to current known treatments. The H1N1 influenza virus has been observed have a rapid mutation cycle that renders it immune to existing drugs. Research equipment to help study the mutation pattern of the virus strain have proven difficult due to the scale of the equipment required to study such behaviors preventing many researchers in pursuing the task.
The aforementioned universities have since then uncovered the key to combat the H1N1 virus mutations using a small computing cluster powered by NVIDIA Tesla GPUs.
Running advanced simulations using the AMBER molecular dynamics application, researchers observed how H1N1 mutations can cause changes in the chemical and biological structure and behavior of a key enzyme of the virus. Armed with this information, they were able to uncover, for the first time, the H1N1 mechanism of resistance to existing anti-influenza drugs. The GPU-based system enabled the team to run larger, more complex simulations which in turn gave them enough time to explore vast mutation possibilities to determine the virus’ immunity. Most importantly, the four-node, eight GPU-based system delivered results in half the time and using one-fifth the servers it would have taken using a CPU-only cluster with 16-24 processors.
Based on the Bristol team’s breakthrough research, it is now possible to identify new ways in which inhibitor drugs can be quickly designed to address future H1N1 mutations, and possibly reduce the deadly impact of future epidemics.  A paper detailing the researchers’ findings has been published in a recent edition of Biochemistry.
University of Bristol
GPUs Help Prevent Next H1N1 Pandemic
Background
A June 2012 report estimates that the 2009 H1N1 influenza pandemic killed more than 284,000 people in the first twelve months the virus was circulating the globe, and possibly as many as 575,400 before it ran its course.
One of the major reasons for the high global mortality and infection rates (89 million people were infected, according to the Centers for Disease Control) was the frequent and rapid onset of virus mutations that rendered existing anti-influenza drugs, such as Tamiflu® (oseltamivir) and Relenza® (zanamivir), ineffective.
Since the 2009 outbreak, researchers worldwide have been working to identify a solution that would allow them to quickly identify new H1N1 mutations, and develop inhibitor drugs to minimize the spread of the deadly virus and prevent the next wide-spread pandemic.
Challenge
Studying viruses such as H1N1 in laboratory experiments is difficult because reactions are often too fast and too fragile to capture. This issue was particularly challenging when studying the H1N1 virus, given its tendency to mutate quickly and frequently.
Supercomputers capable of simulating virus behavior provided scientists with a means to study these H1N1 mutations, but these systems were beyond the reach of many researchers without access to expensive, high-powered supercomputers.
Solution
Researchers at the University of Bristol in the United Kingdom, along with the Bansomdejchaopraya Rajabhat and Chulalongkorn Universities in Bangkok, were able to uncover keys to combating the H1N1 virus mutations using a small computing cluster equipped with NVIDIA® Tesla® GPUs.
Running advanced simulations using the AMBER molecular dynamics application, researchers observed how H1N1 mutations can cause changes in the chemical and biological structure and behavior of a key enzyme of the virus. Armed with this information, they were able to uncover, for the first time, the H1N1 mechanism of resistance to existing anti-influenza drugs.
The GPU-based system enabled the team to run and repeat a much larger number of complex simulations than otherwise would have been possible to perform, and exhaustively explore a multitude of the possible virus mutations to determine what made them resistant to anti-viral drugs.
Most importantly, the four-node, eight GPU-based system delivered results in half the time and using one-fifth the servers it would have taken using a CPU-only cluster with 16-24 processors.
Impact
Previously, the use of computer simulations for drug discovery and disease prevention has been limited because of the large, expensive supercomputers required to study the biological systems. Today a small, affordable GPU-based server gives researchers dedicated access to a high-performance system in-house to power a range of scientific discoveries.
Based on the Bristol team’s breakthrough research, it is now possible to identify new ways in which inhibitor drugs can be quickly designed to address future H1N1 mutations, and possibly reduce the deadly impact of future epidemics.  A paper detailing the researchers’ findings has been published in a recent edition of Biochemistry.
