IBM's Blue Gene Breaks New Research Ground

The four-rack supercomputer system will map protein structures in the hope of manufacturing more effective drugs for humans.
Posted September 7, 2004
By

Clint Boulton


With the interest in using supercomputing to advance medical science still on the rise, IBM said a Japanese research laboratory will use its Blue Gene/L machine to study proteins, hoping to create better drugs to treat human diseases.

Big Blue Tuesday said the Tokyo-based Computational Biology Research Center (CBRC) of The National Institute of Advanced Industrial Science and Technology (AIST) will use Blue Gene/L to map out 3-D protein structures.

IBM declined to list financial terms of the contract. However, such supercomputing pacts generally run in the millions of dollars.

William Pulleyblank, director of exploratory server systems at IBM Research, said his group and AIST are also exploring possible areas for joint research using software IBM has designed to tackle protein simulations on Blue Gene/L.

"We launched the Blue Gene project in 1999. At that time if we look at the challenges of some of these things like protein folding in life sciences, you're going to have to do something different with a computer if you want to attack these, Pulleyblank told internetnews.com. "We set out to build Blue Gene/L."

Pulleyblank said when Blue Gene/L is installed in February 2005, the system will consist of four rack machines, with 1,024 nodes per rack.

It will be based on IBM's Power architecture and will run at a peak processing speed of 22.8 teraflops, or trillions of calculations per second. As of today, this would put it at No. 3 on the Top 500 supercomputing list.

Calling AIST's choice of Blue Gene/L over other systems "validation that we are getting this right," Pulleyblank said the supercomputer will be 24 times more powerful compared to the CBRC's current computer systems.

Moreover, it uses about 1/10th the power per computation of and only about 1/16th the floor space of most systems on the vaunted Top 500 list, where IBM regularly butts heads with such rivals as SGI , Cray , HP , Sun Microsystems and Dell .

The AIST scientists are also creating high-performance computer applications for molecular simulation, mass spectrometry analysis, and cell simulation.

Pulleyblank said that to make Blue Gene effective for disease research, IBM's researchers became well-versed in how genomes work, as well as the complex topic of protein folding.

"What happens with proteins is you get things interacting in odd ways or not behaving properly," Pulleyblank said. "The first thing researchers try to figure out is why, and the second thing they try to figure out is how to intervene to make it stop misbehaving. With machines like Blue Gene we can do that by studying the atoms that make up these molecules and do a custom design of a drug that will respond to a specific mutation of a disease."

IBM Research is getting a lot of mileage out of the Blue Gene project. Pulleyblank said the Argonne National Laboratory in the United States and Dutch astronomical group ASTRON will also be installing Blue Gene/L supercomputers in 2005 to tackle scientific challenges.

Blue Gene/L is also part of the U.S. National Nuclear Security Administration (NNSA)'s Advanced Simulation and Computing (ASC) Program, which IBM has been a partner of since 2001. The NNSA will be installing a very large Blue Gene/L system in 2005 at the Lawrence Livermore National Laboratory to advance understanding of the behavior of materials.






0 Comments (click to add your comment)
Comment and Contribute

 


(Maximum characters: 1200). You have characters left.