If the Human Genome Project gave biologists the equivalent of a parts list for a car, then the latest biology mega-project will be like the instructions that show how the parts combine to make an operating fuel injector. That's one way to describe the aim of a partnership announced yesterday between IBM, the Seattle-based Institute for Systems Biology, the University of Washington and United Devices, an Austin, Texas, software company. The actual goal will be to put together a database of human proteins, the three-dimensional substances that carry out the functions of genes. This collaboration on the Human Proteome Folding Project will be performed on IBM's World Community Grid, which can tap into the computational power of millions of idle computers around the world. The off-hours computer capacity, which can be donated by businesses and individuals, will run calculations to predict the shape of 50,000 to 100,000 proteins that scientists know little about. Once the results are in, they will be available on a public database. Scientists may someday use it to better understand how bodily proteins go bad in diseases like cancer, Alzheimer's and AIDS. Understanding how a protein fold helps researchers know what the protein will or will not do, and how it connects to other proteins. For example, muscle proteins of a certain shape connect to form a muscle fiber. The grid computing technique has been used to sift through data from space for signs of extraterrestrial life, but the protein project will be applied for the first time to biology. When it is complete, the database is expected to hold 400 terabytes of data, which some believe may top the amount now on the Internet. Richard Bonneau, a scientist leading the project at the Institute for Systems Biology, said if enough computer users volunteer from around the world, the millions of calculations to predict protein folding could be done in three to six months.  At the current speed of the institute's own supercomputers, it would take 100,000 years, Bonneau said. "The computational resources are priceless," Bonneau said. Bonneau is a former graduate student at the University of Washington who developed mathematical formulas there to predict protein folding. He said his methods are accurate about half the time, significantly better odds than most who have tried. He will collaborate on the folding project with his former professor at the UW, Dr. David Baker. Bonneau said there will be challenges in storing the data, checking for accuracy and integrating the results with other databases. Then the task is to figure out how to make other biologists use it. "In biology there are no silver bullets, and this is clearly not a silver bullet, but we hope to shine a light on the functioning of 100,000 proteins," Bonneau said. In the words of institute co-founder Leroy Hood, "This project will take an enormous step toward defining the dark universe of proteins." Jennifer Van Brunt, editor of Signals, an online biotech industry magazine, said the project could be very useful for academic and industry scientists. She said companies often struggle to make folded proteins into drugs that bind tightly and properly with cells. But with a deeper knowledge of how proteins configure, she said, scientists could think of ways around the problems. "Ultimately, it will be very useful if you could use it to design proteins from scratch," she said. Source: The Seattle Times