Stimulus funds bring supercomputer to Pittsburgh area
May 5, 2010 4:00 AM
Dr. Markus Dittrich stands in front of a rendition of a photo receptor protein. The biomedical scientist at the National Resource for Biomedical Supercomputing at the Pittsburgh Supercomputing Center based in Oakland will help oversee Anton's operation.
By Sean D. Hamill Pittsburgh Post-Gazette
Starting next fall, the Pittsburgh area will be home to a groundbreaking supercomputer that could help scientists across the country unlock some of the mysteries of life at the molecular level, ultimately helping produce better drugs to treat diseases.
The Anton computer, made by D.E. Shaw Research in New York, is so powerful that the molecular dynamics simulations it can do in a day would take weeks or months on the most powerful supercomputers available today.
Not only that, but it is able to do computations that trace atomic motions over a much longer period of time than ever possible before, allowing scientists to see how critical biological events occur, such as how proteins function, or how a protein and a drug interact, or protein folding. It is believed that diseases can result from improper protein folding.
"This computer does work that really wasn't even possible until now," said Markus Dittrich, a biomedical scientist at the National Resource for Biomedical Supercomputing at the Pittsburgh Supercomputing Center based in Oakland, who will help oversee Anton's operation.
Anton is a massively parallel, 512-node supercomputer, which means it has 512 processors that operate simultaneously, running a seemingly endless series of novel algorithms to project how all the thousands of parts of a protein interact.
The power of a parallel computer is that it can break down sections of a system being studied and spread the analysis to each node. So, if you were, say, studying a system with 51,000 atoms, each node would simulate 100 atoms each.
Other existing computers are typically able to do simulations that show what happens at the molecular level for a few hundred nanoseconds -- a nanosecond is one-billionth of a second -- while Anton is able to measure much longer, to a millisecond -- or one-thousandth of a second -- a critical time period that will reveal much more about protein systems in particular.
Once it is installed over the summer, Anton will be housed at the Supercomputing Center's machine room in Monroeville thanks to a $2.7 million grant from the National Institute of General Medical Sciences, a part of the National Institutes of Health. It was one of 14 Grand Opportunities grants made by NIGMS in 2009 under the American Recovery and Reinvestment Act.
"The Grand Opportunities grants were intended to solicit ideas that can have a big impact on biomedical research, and this fits right into that," Jeremy Berg, director of NIGMS said of the grant for the Anton computer. "This computer has the potential to be a great accelerator in the development of drugs, how drugs work, and how systems work."
The grant will pay for Anton's operation, but the computer itself, which took hundreds of millions of dollars to create over a decade at D.E. Shaw Research, is being loaned to the Supercomputing Center for free for at least a year.
D.E. Shaw Research is a private laboratory founded by David E. Shaw, a legendary former hedge fund manager who used sophisticated computer models for trading on Wall Street in the 1990s to amass a fortune, estimated in 2008 by Fortune magazine at $2.5 billion.
In 2001, he stopped daily work at his hedge fund, D.E. Shaw & Company, to concentrate on Anton and related work at his research laboratory, where he is chief scientist.
The lab dubbed the computer Anton, after Anton van Leeuwenhoek, the 17th century Dutch scientist who is considered the father of microscopy, since it is seen as a "computational microscope" because of its ability to project what is happening in the interaction among particles, atoms and ions.
Famously media-averse, Dr. Shaw did not return an e-mail requesting an interview. A spokeswoman, Jodi Hezky, replied in an e-mail that the company would not comment because it "has been trying to communicate largely through academic talks and peer-reviewed publications in scientific journals, and to avoid attracting attention through non-technical media channels."
Scientists in the molecular dynamics field have known, and been skeptical, about Mr. Shaw's work on Anton for years.
"I was sort of skeptical a little bit when I first heard because I've heard of fairy tales before," said Benoit Roux, a professor of biochemistry and molecular biology at the University of Chicago.
The skepticism was not because they thought it couldn't be done, but because Anton was a leap in speed and power nearly a decade into the future, given the typical pace of computer advancement.
Thomas Cheatham, an associate professor of medicinal chemistry at the University of Utah, was in attendance at an international conference in Barcelona, Spain, in December when Dr. Shaw described just how powerful Anton has proven to be, and he came away, like everyone else, stunned.
"It's a pretty amazing machine," Dr. Cheatham said. But scientists being scientists, "now people would like to get their hands on a machine to see if it can do what he says."
Dr. Roux, who hopes to use Anton when it comes to the Supercomputing Center, said several people who have worked in his lab now work for Shaw "and they tell me it's true what it can do, so, I trust it."
Of course, it is just a machine, and every result has to be confirmed by real lab experiments, which the Shaw company said in its paper that it has done itself and confirmed the accuracy of Anton.
"I always tell my students, 'This is just a model, it's not something real,' " Dr. Cheatham said.
Dr. Dittrich said he expects more than 100 projects to formally apply to use time on Anton during the year or more it will be at the Supercomputing Center.
The $2.7 million grant will not only pay for people to oversee Anton, but to set up a data storage center to store the reams and reams of data that will be generated by the various molecular dynamic simulations.
It was the Shaw company that approached the Supercomputing Center last year with the idea of the loan because of the its international reputation.
Dr. Shaw has "spent a lot of money and expertise to develop these machines and he wants scientists to use it and do good science," Dr. Dittrich said.