Advertisement

Computer simulation yields disease insight

PALO ALTO, Calif., Oct. 20 (UPI) -- By linking together thousands of computers from around the world, scientists said Sunday they have harnessed enough computing power to simulate accurately how proteins fold in the body, an advancement that could lead to a better understanding of and treatments for diseases ranging from Alzheimer's to mad cow.

Protein folding is a process by which proteins assume precise three-dimensional shapes, which are necessary for them to carry out their function in the body. It is critical to understand this process because "a lot of diseases are caused by protein misfolding," Vijay Pande, assistant professor of chemistry and of structural biology at Stanford University and co-author of the research, told United Press International.

Advertisement

"When this process goes wrong a lot of bad things can happen," said Pande, noting protein misfolding may play a role in diseases such as Alzheimer's, Parkinson's and Creutzfeldt-Jakob disease, the human equivalent of mad cow disease.

Unfortunately, protein folding happens so rapidly, in units as small as millionths of a second, it is difficult to study the process in real time. So computer simulations have been suggested as an alternative but up to this point the massive number of calculations involved have proven to be too much for single computers to handle. Using one computer, "it would take 10,000 days or 30 years just to see one protein fold," Pande said.

Advertisement

Pande's team devised a better method by linking more than 30,000 computers together and cutting 30 years down to just 2 years, they report in a study posted on Nature's Web site and soon will appear in a published issue of the British journal.

His group developed a program called folding@home, which divides up the vast amounts of calculations for studying protein folding and distributes it to thousands of computers around the world -- much the same way that thousands of personal computers are helping to process possible signals received from alien civilizations in the Search for Extraterrestrial Intelligence, or SETI, at home program.

The folding@home program ran as a screensaver or as a low priority program in the background on the desktop computers of people who downloaded it voluntarily. The program essentially was invisible to the volunteers, Pande said. This technique is known as distributed computing and the 30,000 computers linked together created slightly more computing power than all of the supercomputers operated by the National Science Foundation combined, he said. This allowed them to create simulations that are "1 million times longer than typical simulations do these days."

Using a simple protein that Martin Gruebele, a professor of chemistry, physics and biophysics at the University of Illinois at Urbana-Champaign, had studied extensively in his lab, Pande's team found the computer simulation of its folding agreed very well with experimental observations. It is an accurate and dependable simulation and could be useful for designing new drugs based on protein folding, Pande said. He noted a number of pharmaceutical companies already have expressed interest in the technology.

Advertisement

"This gives us further indication that simulation methods will be able to address questions related to protein folding and probably more importantly protein misfolding" as it relates to disease, Charles Brooks, a professor of biophysics at Scripps Research Institute, who wrote a commentary on the research in Nature, told UPI. "This paints a bright picture for using simulation in those areas of biomedical interest," he added.

Pande agreed that distributed computing holds great promise for studying protein folding. "It lets us get our hands in the system, it lets us be almost like a god in that world," he said, adding that the number of volunteers now is up to 40,000 and about 300 new users download the program to their computers every day. This gives Pande's team computing power that "is 100 times what our colleagues have, so I see a greater burden on us to really do something grand," he said. "If we can't do something that other people haven't done, we've done something wrong."

Pande said his team will "now apply this to Alzheimer's and attempt to understand the nature of protein misfolding in that disease." There is no cure for Alzheimer's, which currently affects 4 million Americans and could rise to 14 million by 2050 if no treatment is found.

Advertisement


(Reported by Steve Mitchell, UPI Medical Correspondent, in Washington)

Latest Headlines