Supercomputer Provides Simulations for Alzheimer's Study

For the past five years, Joan-Emma Shea, a professor at the University of California, Santa Barbara, has run thousands of amlyoid peptide simulations on the Ranger supercomputer at the Texas Advanced Computing Center (TACC) in an attempt to better understand Alzheimer's disease.

Shea is using the 579 teraflop supercomputer to look for data in support of the hypothesis that says toxcity in the brain is cause by small, transient molecules.

While the accumulation of amyloid plaques ‭ long knotty fibrils that result from misfolded proteins ‭ has long been associated with brain cell death, researchers have now begun to look at oligomers, the precursors of fibrils.

Her work is made possible with a grant from the National Institutes of Health and the National Science Foundation-funded Extreme Science and Engineering Discovery Environment (XSEDE) initiative, which aims to make computational resources widely available to the US research community.

Relatively speaking, a 579 teraflop supercomputer is not that impressive in a world where the fastest supercomputer currently clocks in at 16.32 petaflops. But this goes to show that the XSEDE program has the right stance in that researchers don't need the fanciest, fastest system in the world ‭ they just need easy access to a reliable HPC system with a respectable flops performance.

The Ranger supercomputer allows Shea to model or simulate what structures the amyloid peptides are adopting, at resolutions far exceeding what is possible experimentally.

"The number of atoms is huge•we need a lot of computational resources to simulate them. Nothing that we're doing here is something that we could do on our home clusters. The scale of it is intractable," says Shea. "With growing computational resources and capabilities, we'll be able to look at how these proteins interact with membranes...We're far away from simulating a whole cell, but we can start incorporating additional elements that may turn out to be important."

Ranger is a Linux-based system with eighty-two compute racks housing the quad-socket compute infrastructure that uses the Lustre file system across 72 I/O servers.

Soon, Shea will be able to take advantage of the TACC's "Stampede" supercomputer, which is slated to come online in early 2013 and will be 20 times more powerful than Ranger.

To submit a proposal to request an allocation, you can visit the XSEDE website.

Matthew Dublin is a senior writer at Genome Technology.