Dr. Michael Giddings
Macs and the Next Frontier: Proteomics
Using the Power of Xserve
To scan proteins in yeast, Dr. Giddings relies on a single Power Mac. But for the human genome experiments, he needed a much larger resource. So he purchased a 42-node cluster of Xserve servers.
“With the Xserve,” he explains, “we have a lot more power to bring to bear on our research. Now, running a scan against the whole human genome will take about 16 minutes if it’s distributed over the whole cluster,” he says. “If we had to do it on a single machine, it would take a couple of days.”
Dr. Giddings started writing the UNIX application on a Mac while he was doing post-doctoral research at the University of Utah. He chose Macs because “I very much like working in an environment where I have the UNIX command line, combined with a lot of great programming tools, combined with all of my desktop conveniences — everything from Microsoft Office to the mail program. I like the simplicity combined with the power.”
He also is comfortable administering a cluster of Macs. “Some of the other machines we considered,” says Dr. Giddings, “required a lot more dedicated training and support staff. We considered Linux machines and we could have built up a big cluster of those, but I have talked to a number of different people who have had a fairly high hardware failure rate and also a lot of software hassles in putting those together and maintaining them.”
“We wanted a plug-and-play solution that was still cost effective yet could meet our computing needs. That’s the place that Xserve and Apple fits in pretty well, especially with the new cluster nodes.”
The (Mostly) Mac Lab
Other than two PCs that are “used occasionally for certain software,” says Dr. Giddings, everybody else in his lab of nine researchers uses Macs.
“Running a scan against the whole human genome will take about 16 minutes if it’s distributed over the whole Xserve cluster.”
“It’s kind of interesting,” he observes. “A few people who joined my group are programmers who had a lot of experience with Windows. They came in and started working on Windows — I didn’t try to force them into going with Mac OS X — but all of them have converted and are now using Macs.”
Moving Toward Cell Modeling
Dr. Giddings expects to get good results from his research — to know what’s changing and why — within two years. But that’s just the beginning. “There haven’t been technologies in the past,” he says, “that have been easily able to monitor genomic changes as they occur rapidly in a population. Our application and the Macs offers a solution for that, and we’re very excited about it.”
As for the future, Dr. Giddings predicts that multiple areas of bioinformatics will ultimately converge into simulation and cell modeling. “When we get to the point where a lot of people are doing this, the computing resources that are going to be needed are tremendous. So the need for good, cost-effective computing resources will continue to grow.”