By Salvatore Salamone
August 15, 2003 | The Scripps Research Institute (TSRI) is rolling out phase two of its FightAIDS@Home project which, unlike phase one, will tackle the high mutation rate of HIV.
“HIV is a very slippery target,” says Arthur Olson, director of the molecular graphics laboratory at TSRI, not because the virus is intelligent, but because HIV has a “sloppy way of copying its RNA from one generation to the next.” Researchers estimate one error is made per every 10,000 to 100,000 HIV bases replicated, which produces about one mutation every time a virus is replicated.
During phase one of FightAIDS@Home -- a proof-of-concept effort begun in September 2000 -- roughly 60,000 computers in 20 countries collectively logged 1,400 years of computing time and performed more than 9 million molecular docking calculations. Distributed computing vendor Entropia helped to work out the technical details, and a library of candidate inhibitor molecules was tested against a single target.
“In phase one, we were able to demonstrate the feasibility of the project,” Olson says. That work combined expertise in several areas, including virology, X-ray crystallography, and computational chemistry, and it produced new computational methods to look at drug resistance. “We wanted to know how to design [drugs] with knowledge of drug resistance,” Olson says.
At the heart of phase two is computer simulation of compound docking in which candidate molecules (inhibitors) are tested against not one target, but possibly millions of mutated variants of the HIV protein.
“This is a much larger project than going after a single target,” Olson says. “We’ll have a large matrix with many inhibitors that need to be checked against multiple mutations.” One challenge for TSRI will be to use the donated computer resources efficiently. The computational load will grow in proportion to the number of target mutations.
The lab uses a program called AutoDock to predict how drug candidate molecules might bind to an HIV target receptor. The program models atomic and molecular forces to predict how a molecule binds to a target. Essentially, the program is looking for molecules that can block certain proteins from binding, thus disrupting an HIV pathway.
Recent research has dramatized the challenge of such an undertaking. In March, a group led by Douglas Richman, a virologist at the University of California, San Diego School of Medicine, published a paper in Proceedings of the National Academy of Sciences that looked at the HIV mutation process. In the paper, Richman noted that while HIV-infected patients quickly develop an antibody response against the virus, the same antibodies “force [HIV’s] ongoing evolution into new strains that dance around the antibody response.”
“The neutralizing antibodies are exerting a selective pressure on the virus, and the virus is continually mutating to avoid it,” Richman says. His team found that antibodies in an HIV-infected person change their activity to try to keep pace with the changing virus. But HIV evolves faster than the antibody response.
“The bad news is that the HIV virus is always a step ahead, and the antibody response can’t control it,” Richman says. Antibodies hold promise as a therapy or vaccine, he concludes, but only if they can be engineered to recognize many different strains of HIV.
FightAIDS@Home hopes to quickly incorporate new knowledge, such as that in the recent Richman paper. Scripps has also decided to bring management of the distributed application completely in-house rather than rely on Entropia.
“With traditional drug design, you would take a target and throw a bunch of [molecules] at it to see what fits best,” Olson says. This common approach to drug discovery -- a mainstay in traditional computational biology done by many life science companies -- has also been used in other distributed computing efforts to date. For instance, last year Oxford University conducted an anthrax research project, during which volunteered computers screened 3.5 billion molecules against a single target (see “Computing a Cure for Anthrax,” March 2002 Bio-IT World, page 12).
“We would like to know which drug or combination of drugs will box [HIV] into a corner,” Olson says, adding that his team will rely on the work of other scientists, such as virologists and X-ray crystallographers, who will continuously find new variants of HIV to be used as targets.
Initially, the project will rely on the scientists themselves to guide the computational work. “We’ll try the best inhibitors against [specific] mutations,” Olson says. But as the project progresses, TSRI will bring in other technology to help narrow the choices and thus make more efficient use of the computational resources.
He notes that future FightAIDS@Home work will likely use a learning system to examine computational results and automatically hone in on the best combination of inhibitors to test against certain targets.
To participate in phase two of FightAIDS@Home, visit the Scripps home page (fightaidsathome.scripps.edu) and download the necessary software. Once the software is downloaded, it runs in the background whenever the computer is turned on. Essentially, the software includes an executable code that runs AutoDock on the PC, and a small set of molecules or compounds to test against a target.
Whenever a home computer running the software connects to the Internet, the application checks in with a server program running at Scripps. If all the test molecules and compounds have been run through the AutoDock program, the findings are passed along to a central database that tracks the results of all the volunteer efforts.