Bilingual Expertise

TACC scientists work alongside domain researchers to tackle major computational problems

The crucial ingredients for running a high-performance computing center are: powerful supercomputers; comprehensive software suites; and experts who provide assistance when users need help. But once these basics have been provided, what’s the next step to advance research?

“You’d like to foster an environment where you have domain experts on staff who are also consumers of the supercomputing products that we’re providing as a center,” said Karl Schulz, director of the applications collaborations group at the Texas Advanced Computing Center (TACC). “That’s the foundation for collaborative groups like ours.”

Many of TACC’s HPC technology experts also have know-how in biology, engineering, aeronautics and other fields of science. This specialized knowledge allows them to participate at a more fundamental level with key research projects on which they collaborate.

“We’re attempting to inject some of the collective expertise that TACC has a little bit earlier into the research process,” Schulz said. “By having staff work directly with researchers, these collaborations go beyond the scope of what we can do in consulting. We can affect the development of the codes so they map well onto the types of systems which we’re deploying now and which will be deployed in the future.”

Schulz, a computational engineering scientist, and Michael Gonzales, computational biology program director, are spearheading the application collaboration effort at TACC. In the coming year, the Center expects to hire lead collaborators for energy- and geosciences-related research as well. With extensive knowledge of HPC best practices and engineering and biology, Schulz and Gonzales are well positioned to accelerate numerous projects in their field.

“You can get over the hump a lot quicker if you can embed someone who speaks both languages,” Schulz said.

Schulz is currently working with The University of Texas at Austin’s Institute for Computational Engineering and Science (ICES) on a Department of Energy project to advance the science of uncertainty quantification applied to the modeling of atmospheric reentry. The Center for Predictive Engineering and Computational Sciences (PECOS) brings together an interdisciplinary, multi-university team to develop the next generation of advanced computational methods for predictive multiscale, multiphysics simulations.

Since life science is a relative newcomer to HPC, Gonzales’ collaborative efforts have been directed in a slightly different vein. Before embarking on major collaborations, Gonzales gathered a full suite of domain-specific software and tailored these tools to run optimally on Ranger’s architecture. He also developed and delivered a series of life science-specific computational training workshops targeted at providing life science researchers with the tools needed to make the most effective use of TACC’s resources.

“I’m an experimental biologist by training, but over the last decade I’ve been doing computational biology,” Gonzales said. “My background enables me to better understand the questions researchers are asking and know how to apply supercomputing technologies to these questions—it’s a very unique perspective.”

In addition to improving TACC’s bio-cyberinfrastructure generally, Gonzales has embarked on a number of direct collaborations that push the boundaries of computational biology. Working with Scott Stevens from the University of Texas at Austin, Gonzales is assembling complete genome of an extremophile bacterium found in deep-sea vents. The work is a great example of the combination of computational and experimental that these collaborations bring. Ultimately, they hope the results will help scientists understand how the C. caldarium’s proteins, so similar to our own, can withstand temperatures up to 600°F whereas human proteins melt at 110°F.

Another collaboration with the Texas Institute for Drug and Diagnostic Development (TI-3D) and Dr. Pengyu Ren in the Department of Biomedical Engineering uses the power of Ranger to produce more accurate drug docking measurements and to predict which molecules might act as drug compounds. Docking is a popular computational approach for screening potential drug compounds. The advantage of docking simulations is their low cost and speed, which makes it possible to screen of millions of compounds. Although these algorithms can reliably predict the correct binding mode, the current methods are not able to consistently identify a compound’s efficacy. This shortcoming greatly reduces its practical use. “We believe that by leveraging current computational capabilities, we will be able to take a more comprehensive approach and construct a more realistic and accurate model of protein-ligand interactions,” Ren said.

All of these projects benefit from Schulz and Gonzales’ familiarity with best computational practices in their science domain and their ability to take advantage of Ranger’s capacity and capabilities.

“The goal of these collaborations is two-fold,” Gonzales said. “One is to create new means to address intellectually stimulating and scientifically important questions. But the follow-on is to take that knowledge—or the software, or whatever is developed on the computational side—and make that part of TACC’s infrastructure to provide it to the community as a whole.”

There is also a third need served by the creation of the application collaboration group. Through their collaborations, Schulz and Gonzales are able to experience TACC’s systems, infrastructure and institutional practices as an outside user would. They can then communicate their experience as consumers of TACC’s systems back to TACC’s administrators.

“We recognize the importance of obtaining the best user feedback on our technology resources and support services,” said TACC Director Jay Boisseau. “The insights we get by participating in these collaborations improves the overall infrastructure and support services, and helps communities advance their research faster and farther.”

For Schulz and Gonzales, their work leverages the knowledge they’ve accumulated over decades to help the most promising research projects do their most effective work.

“After spending a number of years in the supercomputing environment, I’ve been fortunate to learn an incredible amount about how to use HPC systems effectively, how to eke out extra performance, and how to do quality software engineering,” said Schulz. “It’s nice to have an opportunity to be involved directly in the development phase and to contribute to the long-term design of a particular application.”

Aaron Dubrow
Texas Advanced Computing Center
Science and Technology Writer
September 29, 2009

The Ranger supercomputer is funded through the National Science Foundation (NSF) Office of Cyberinfrastructure “Path to Petascale” program. The system is a collaboration among the Texas Advanced Computing Center (TACC), The University of Texas at Austin’s Institute for Computational Engineering and Science (ICES), Sun Microsystems, Advanced Micro Devices, Arizona State University, and Cornell University. The Ranger and Lonestar supercomputers, and the Spur HPC visualization resource, are key systems of the NSF TeraGrid (www teragrid.org), a nationwide network of academic HPC centers, sponsored by the NSF Office of Cyberinfrastructure, which provides scientists and researchers access to large-scale computing, networking, data-analysis and visualization resources and expertise.