Rice University breaks down barriers to research

Founded in 1912, Rice University (Rice) is a leading university in Houston, Texas. Committed to undergraduate education, Rice is one of the top research institutions in the United States and works in partnership with several local medical institutions. In 2010, Rice opened the BioScience Research Collaborative (BRC) to provide scientists and educators from Rice and its longtime partners in the the Texas Medical Center a space dedicated to collaborative research.

According to Kamran M. Khan, vice provost for information technology at Rice University, “The BRC is designed to facilitate and encourage interdisciplinary interactions between researchers and is equipped for cutting-edge laboratory, theoretical and computational investigations.” Biomedical research projects underway at the BRC include protein folding analytics, artificial organ implant modeling, multispecies genome mapping, biomolecular structural modeling and drug-drug interaction analytics. The university believes that advancing research in healthcare and related fields will ultimately lead to finding cures that will benefit humankind.

Research of this nature often requires intense computing power—power that Rice was not always able to deliver. “We would hear repeatedly from researchers that the systems were just not able to handle the workloads the researchers required,” says Jan Odegard, executive director, Ken Kennedy Institute for Information Technology at Rice University. Some of the programs frequently used by researchers can take days or even weeks to run. The university’s systems also lacked the memory and storage capacity researchers required. “Our previous clusters had limited on-node memory and very little storage for persistent data,” says Kim Andrews, manager of research computing, Rice University. “These limitations constrained the type of research we could support and presented an obstacle to many natural science researchers, forcing them to look outside the university or to supercomputing centers to meet their needs.”

A deepening partnership with IBM
“Over the past few years, Rice and IBM have enjoyed a productive, collaborative relationship, working together to make a difference in the world,” says Khan. The two organizations have continued to discuss opportunities to work together on innovative projects, since Rice and IBM share a vision for the future of computing-assisted research. Removing computer-related constraints from researchers can pull computing limitations out of the equation and make a positive impact for research. “Getting science done is the target, whether it is computational or hands-on in a lab,” says Andrews.

In December 2009, IBM awarded Rice the IBM Shared University Research (SUR) award. Sponsored by the IBM Systems and Technology Group University Alliances and IBM University Relations teams, the SUR grant included the donation of the BlueBioU 19-node high-performance computing (HPC) environment. Based on IBM POWER7 technology, the BlueBioU platform is as powerful as the combined total of the university’s previous supercomputers. IBM also committed to a multiyear alliance with Rice to help configure and tune the HPC environment. “Partnership with IBM enables both Rice and IBM to work collaboratively on current and future problems facing our planet. Sharing knowledge and bringing researchers together provide powerful synergies to focus and work on common goals,” says Khan.

Taking collaboration to new levels
Numerous groups within IBM worked with Rice to make the BlueBioU project a success. “The help we got from the IBM teams was just incredible. It would have been more challenging to go and do this by ourselves,” says Andrews. Because Rice had been using x86 technology-based platforms for many years, it needed help understanding the POWER7 architecture and how to integrate it into its existing computing environment. The IBM Linux Technology Center helped Rice benchmark the new system, integrate it with the university’s existing storage systems and take advantage of the IBM compilers. The Linux team, in turn, collaborated with the IBM compiler development team, which provided crucial information to help Rice optimize performance using the IBM XL compilers.

Rice also tapped into the knowledge and experience of the IBM Worldwide Deep Computing technical team. Scientists and specialists from that group worked with prospective users of the BlueBioU system to help Rice develop a work plan for the research that would be performed on the system during its first year. The Worldwide Deep Computing team also helped Rice compile and install software on the BlueBioU system and tuned code that enabled researchers to benefit from the high performance provided by IBM compilers. “The SUR award wasn’t about a company coming in and simply dropping off hardware. We spent hundreds and hundreds of hours collaborating with the IBM teams,” says Khan. “I’ve rarely seen this
level of collaboration before.”

Harnessing the potential of the POWER7 platform
Many of the applications used by biomedical researchers are extremely resource-intensive. In some cases, researchers are actually creating an alternate reality within the computer. “We try to create a surrogate reality that mimics true reality,” says Andrews. “We pose questions to the surrogate reality, and we get answers that we hope are similar to what we’d get in reality.” The more detailed that surrogate reality is, the more accurate the results of the research will be.

The key to creating this degree of detail is sufficient online memory that can store information and ample processing power to manipulate it rapidly. The BlueBioU system provides the necessary memory and power to support these realities. By using the BlueBioU system, for example, researchers can store multiple copies of the human genome database in memory rather than having to rely on an external copy, which enables them to run computations on it three times faster than they could in the past. But it’s not just about faster computations; it’s about real-world results. Geophysicists at Rice are currently using the system to develop methods for predicting tsunamis and anticipating how toxins might spread through a water source—studies that could one day help save lives.

Another example of the power of the BlueBioU system is illustrated by the work on protein folding performed by Jianpeng Ma, professor of biochemistry and molecular biology at Baylor College of Medicine and professor of bioengineering at Rice University. Between the greater memory of the BlueBioU system and the ability to perform the work on-site at Rice, Ma was able to perform his calculations 26 times faster than he could in the past. That’s the difference between waiting a month for results and waiting a single day. His work may one day be used to help understand and prevent many of the diseases caused by protein that folds incorrectly, such as Alzheimer’s, cystic fibrosis, emphysema and certain
types of cancer.

Enabling researchers to focus on the research,not on IT
Not only has the BlueBioU system provided more computing power for Rice researchers, but it has also relieved them from having to worry about IT. In the past, many faculty members and researchers maintained what were essentially mini data centers for their departments, stored in back rooms or closets and maintained by graduate students. When Rice obtained the BlueBioU system, it hoped to convince researchers to move their workloads to the new system. Because many researchers work with the same applications for years and years, they understandably had some reservations about moving these applications to a new platform and the effect that this move would have on their work.

The POWER7 architecture produced such powerful results that it was easy to convince researchers to give up their private servers and instead take advantage of the BlueBioU system. And with help from the IBM teams, researchers were able to smoothly port their applications to the POWER7 platform. The IBM XL compilers proved to be key in these efforts. “As soon as you start tuning the IBM compilers, you get better performance from just about any application,” says Andrews. “Researchers would rather not spend time tweaking the code. They can let the compiler do it for them.” “Our faculty and researchers now have access to a powerful, high performance computing environment that translates to faster results, optimized code and the flexibility to run multiple jobs in parallel,” says Odegard.

Expanding the system—along with the possibilities
Because the BlueBioU solution enabled incredible successes for the faculty and researchers from Rice and the Texas Medical Center, Rice received a grant from the National Institutes of Health (NIH) to expand the computing power, storage and networking capabilities of the BlueBioU system in support of bio-computational research. The system grew from 19 nodes to 49 nodes and gained 270 TB of usable disk space. To further collaboration with external organizations as well as with the rest of the Rice campus, Rice upgraded the networking capabilities to increase bandwidth to 10 Gigabit Ethernet. “The early exposure and success with the POWER7 platform was a critical factor in Rice selecting IBM and POWER7 as the architecture for the NIH award it received in 2010,” says Odegard. “This success enabled Rice to expand the cluster from 19 to 49 nodes.”

As the system grows, there’s really no limit to what the faculty and researchers who use it can accomplish. “The SUR Award has enabled Rice to provide additional computational power to the faculty and researchers doing cutting-edge research that will have a great impact on biomedical fields and ultimately lead to discoveries for the betterment of human life on our planet,” says Khan.

For more information
To learn more about the POWER7 platform from IBM, please contact your IBM marketing representative or IBM Business Partner, or visit the following website: ibm.com/systems/power/

Additionally, financing solutions from IBM Global Financing can enable effective cash management, protection from technology obsolescence, improved total cost of ownership and return on investment. Also, IBM Global Asset Recovery Services help address environmental concerns with new, more energy-efficient solutions. For more information on IBM Global Financing, visit: ibm.com/financing