GNN - Genome News Network  
  Home | About | Topics
Global network of personal computers simulates protein folding

A global network of personal computers dedicated to solving biological problems has simulated one of nature's greatest tricks: The efficient folding of a human protein. Scientists have used the network to predict the behavior of a modified protein and confirmed the results experimentally.

The findings, reported in Nature, demonstrate a principle. They show that networks of computers can accurately simulate the dynamics of a small protein folding. The shapes of proteins determine their functions in cells, and misshapen proteins are thought to contribute to a number of diseases, including Alzheimer's.

Detail from diagram of protein folding. View full

Understanding how proteins behave could lead to new drugs that target errors in these vital molecules, as well as to synthetic therapeutic substances that don't exist in nature.

The immediate value of the study is that it gives scientists confidence in the ability of computers to simulate protein folding, says Martin Gruebele, a biochemist at the University of Illinois, Urbana, who co-led the research.

"In the end," says Gruebele, "we want to be able to tell researchers that for any DNA sequence that encodes a protein we can calculate what that structure looks like without having to generate the protein itself."

"But," he adds, "this will only work if you have trust in the calculations."

The molecule in the study was a slightly modified form of a protein called BBA5, which helps cells utilize zinc. To check the accuracy of the computer methods, a team of biochemists analyzed the protein experimentally. They added a naturally fluorescing amino acid to make the protein visible as it folded.

The computer predictions were in agreement with the calculations made later experimentally. The agreement was "almost too good," says Gruebele, adding that a larger difference between the measurements would have been satisfactory.

Working from opposite directions, the computer group obtained a folding time of between 6 and 8 microseconds, while the biochemists put the figure at about 7.5 microseconds. A microsecond is a millionth of a second.

The two groups were also close in their measures of the stability of BBA5, another key factor for proteins.

The study is based on projects led by Vijay Pande, a physical chemist at Stanford University in California and co-author of the Nature paper. The projects, Genome@home and Folding@Home, are experiments in using networks of personal computers to solve massive calculations.

More than 30,000 people and their home computers participate in these projects every day. So far the researchers have focused their efforts on small proteins consisting of fewer than 50 amino acids. But their goal is to tackle proteins that have 100 amino acids, roughly the number needed to study human diseases that involve folding problems.

To reach that goal may require another 10-fold increase in computing power, or another 270,000 volunteers willing to surrender their screen savers to science.

The strategy of 'distributed computing' was pioneered by the Search for Extraterrestrial Intelligence Institute in Mountain View, California, known as SETI. For several years the group has searched for electronic transmissions from alien life forms.

See related GNN article
»Genome@home: Mass-computing effort brings public into genome fold

. . .

Snow, C.D. et al. Absolute comparison of simulated and experimental protein-folding dynamics. Nature. Published online October 20, 2002.

Back to GNN Home Page