Press Archive - 2000-Sep, Wired News

Protein Fiends Join the Fold

Wired News
Sep. 26, 2000
by Andy Patrizio

With the human genome now mapped, the freeway on-ramp to the better understanding of life has been opened.

But there's a leap between knowing all the genes, and knowing how proteins are formed, take shape, and behave.

One thing scientists do know is that proteins do something unusual in biology: They self-assemble. It's like a car building itself on the assembly line. This self-assembly is called "folding."

Scientists also know that if they are to learn what they need to learn as quickly as possible, they're going to have to enlist the help of computer-savvy volunteers.

To better understand how proteins fold in the hopes of discovering possible cures for illnesses, Stanford University's Pande Group is hoping for the help of the masses by taking the same approach as SETI@Home, Entropia, and Distributed Sciences: It's seeking Internet users willing to put their idle PCs to work running a massively distributed simulation.

Only a week old, the Folding@home program is designed to run simulations of how proteins assemble themselves. The program can be used on PCs running Windows or Linux.

Cracking the assembly process is a Holy Grail for biochemists.

"Once you know what the shape is, you can guess at what it does," said Dr. Vijay Pande, assistant professor of chemistry at Stanford University and principal investigator on the project. "Their structure tells us a lot of what they do and how they put themselves together."

Antibodies that help fight viruses have a particular structure, and that shape attaches onto the virus to fight it. By understanding the shape, scientists can understand how antibodies do their job.

The goal, Pande said, is to understand how proteins assemble themselves and what the final structure is. While we know of many proteins involved in illness, biochemistry, and so forth, we don't know how they get into that state.

By doing these folding simulations, we can understand which genes activate particular proteins, and perhaps how to treat diseases that emanate from the dangerous and deadly ones, he said.

Folding happens at an extremely rapid pace, often in a matter of microseconds. However, each step of the folding process is enormously complex, and computers simply can't calculate all of the variables.

There are so many variables and calculations to be done in the space of nanoseconds that the PC simply can't keep up, according to Pande.

Simple proteins assemble in about 10,000 nanoseconds, and one 400MHz computer can simulate 1 nanosecond of assembly time in about a day, Pande said. This means 1,000 participants would be needed to simulate a simple protein fold in 10 days, and that's just for simple folds. More complex proteins can take far longer.

The Pande Group used a software developer's kit for building its service, called the Mithral Client-Server SDK.

The service is designed to quickly build a distributed-computing and peer-to-peer network, to whip out your own SETI@Home-like project.

Pande, a scientist, didn't want to be consumed with the computing side of it. Now that researchers know the Mithral SDK, they can build a client in a week or two, he said.

Pande can't discuss some details of how Folding@home works or what it has simulated, mostly because some of it is currently under scientific peer review, and partly because it's only been officially running for a week. But all results will be open and published.

Because genes are the blueprint for forming proteins, just knowing the human genome sequence is like having a list of all the parts in your car. It doesn't do you much good without knowing how they are assembled, and then knowing what it does once assembled.

Proteins are the biological building blocks for actions in an organism. As enzymes, they drive biochemical reactions in the body, such as digestion. As antibodies, they recognize disease and other foreign bodies in the organism and help the immune system eliminate these intruders.