For the first time, a distributed computing experiment has produced significant results that have been published in a scientific journal. Writing in the online edition of Nature magazine, Stanford University scientists describe how they — with the help of 30,000 personal computers — successfully simulated part of the complex folding process that a typical protein molecule undergoes to achieve its unique, three-dimensional shape.From Stanford University:
Folding@home scientists report first distributed computing success
As you read this sentence, millions of personal computers around the world are working overtime — performing complex computations on their screensavers in the name of science. This growing Internet phenomenon, known as “distributed computing,” is being used for everything from the search for extraterrestrial intelligence to the design of new therapeutic drugs.
Now, for the first time, a distributed computing experiment has produced significant results that have been published in a scientific journal. Writing in the advanced online edition of Nature magazine, Stanford University scientists Christopher D. Snow and Vijay S. Pande describe how they — with the help of 30,000 personal computers — successfully simulated part of the complex folding process that a typical protein molecule undergoes to achieve its unique, three-dimensional shape. Their findings were confirmed in the laboratory of Houbi Nguyen and Martin Gruebele, scientists from the University of Illinois at Urbana-Champaign who co-authored the Nature study.
Every protein molecule consists of a chain of amino acids that must assume a specific three-dimensional shape to function normally.
“The process of protein folding remains a mystery,” said Pande, assistant professor of chemistry and of structural biology at Stanford. “When proteins misfold, they sometimes clump together, forming aggregates in the brain that have been observed in patients with Alzheimer’s, Parkinson’s and other diseases.”
How proteins fold into their ideal conformation is a question that has tantalized scientists for decades. To solve the problem, researchers have turned to computer simulation — a process that requires an enormous amount of computing power.
“One reason that protein folding is so difficult to simulate is that it occurs amazingly fast,” Pande explained. “Small proteins have been shown to fold in a timescale of microseconds [millionths of a second], but it takes the average computer one day just to do a one-nanosecond [billionth-of-a-second] folding simulation.”
Simulating protein folding is often considered a “holy grail” of computational biology, he added. “This is an area of hot competition that includes a number of heavyweights, such as IBM’s $100 million, million-processor Blue Gene supercomputer project.”
Two years ago, Pande launched Folding@home — a distributed computing project that so far has enlisted the aid of more than 200,000 PC owners, whose screensavers are dedicated to simulating the protein-folding process.
The Stanford project operates on principles similar to earlier projects, such as SETI@home, which allows anyone with an Internet connection to search for intelligent life in the universe by downloading special data-analysis software. When a SETI@home screensaver is activated, the PC begins processing packets of radio telescope data. Completed packets are sent back to a central computer, and new ones are assigned automatically.
For the Nature study, Pande and Snow, a biophysics graduate student, asked volunteer PCs to resolve the folding dynamics of two mutant forms of a tiny protein called BBA5. Each computer was assigned a specific simulation pattern based on its speed.
With 30,000 computers at their disposal, Pande and Snow were able to perform 32,500 folding simulations and accumulate 700 microseconds of folding data. These simulations tested the folding rate of the protein on a 5-, 10- and 20-nanosecond timescale under different temperatures. Using these data, the scientists were able to predict the folding rate and trajectory of the “average” molecule.
To confirm their predictions, the Stanford team asked Gruebele and Nguyen to conduct “laser temperature-jump experiments” at their Illinois lab. In this technique, an unfolded protein is pulsed with a laser, which heats the molecule just enough to cause it to bend into its native state. A fluorescent amino acid imbedded inside the molecule grows dimmer as the protein folds. Researchers use the changing fluorescence to measure folding events as they occur.
The results of the laser experiments were in “excellent agreement” with the Folding@home predictions, Pande and his colleagues concluded. Specifically, the computers predicted that one experimental protein would fold in 6 microseconds, while laboratory observations revealed an actual folding time of 7.5 microseconds.
“These experiments represent a great success for distributed computing,” Pande said. “Understanding how proteins fold will likely have a great impact on understanding a wide range of diseases.”
The Nature study was supported by the National Institutes of Health, the American Chemical Society, Intel and the Howard Hughes Medical Institute.