Supercomputer Tackles Mystery of HIV's Structure
The researchers used the supercomputer Blue Waters to determine the complete HIV capsid structure, a simulation that accounted for the interactions of 64 million atoms.
CREDIT: Klaus Schulten/Juan Perilla, Beckman Institute
Using a simulation of more than 64 million atoms and a supercomputer named Blue Waters, researchers at the University of Illinois have created a digital model that could provide the key to curing HIV.
The model is of the HIV particle's capsid, which is an inner protein shell that protects the virus's genetic material, a DNA-like molecule called RNA.
When the HIV particle enters a human cell such as a white blood cell, however, the capsid has to open to release the RNA, which then goes on to infect the host cell's DNA. [See also: What is HIV & AIDS?]
Researchers also know that monkeys, who are immune to HIV/AIDS, don't become infected because they can attack and shut down the virus's capsid, though it's not yet known how this is accomplished. All this makes the capsid an ideal target for scientists trying to develop medication to combat the virus.
But scientists are unable to get a clear view of the HIV capsid in its entirety using normal observational techniques — they are able to get either highly detailed looks at small parts of the capsid, or low-resolution looks of the assembly as a whole.
Further complicating the study is the fact that the HIV capsid is polymorphic, meaning it's constantly changing its shape. Scientists were able to determine that the capsule is comprised of proteins in hexagonal and pentagonal patterns, but its polymorphic nature means that observing single instances of capsid structure isn't very useful.
[See also: Baby with HIV is Cured, Doctors Say]
The challenge of creating a functional simulation of the HIV capsid is comparable to trying to replicate a fully functional video game based on a few screenshots of gameplay. Those screenshots are incidental results of the game's underlying code, not information on the code itself. Similarly, scientists needed a way to determine the rules governing the capsule's polymorphic, changeable nature.
University of Illinois physics professor Klaus Schulten has been collecting data on HIV capsids for decades. But synthesizing this information into a detailed computational model — consisting of more than 64 million individual atoms, as well as all the known behavioral patterns and relationships between them — would require an enormous amount of processing power.
That processing power presented itself to Schulten and his team in the form of Blue Waters, a supercomputer developed by the University of Illinois' National Center for Supercomputing Applications. Blue Waters is a petascale computer, meaning it can perform more than one petaflop (a quadrillion operations) per second. It's housed in an 88,000-square-foot facility on the university campus and needs a complex cooling system to prevent overheating.
Using Blue Waters, Schulten and his team were able to synthesize vast amounts of data on the HIV capsid, its molecules, and known and projected behaviors. The computer then put it all together, creating a dynamic and highly detailed model via a complex methodology that Schulten dubbed "molecular dynamic flexible fitting."
It took Blue Waters a full month to run the program. But the process was a success: the first complete HIV capsid structure, consisting of 216 protein hexagons and 12 protein pentagons, all joined together at varied angles. It's the largest computer simulation ever run, the researchers say.
The work's not done yet, though: The next step is to use this highly detailed model to try to figure out how the same protein, in a few basic shapes, is capable of such malleability even within the same capsid.
"This capsid model … is the platform for the development of new therapies," said postdoctoral researcher Juan Perilla in a video explaining the team's findings. "And that's why it's so important."