PTI | Research | Scholarly Highlights

From palm leaves to pixels: TeraGrid’s Data Capacitor supports digital preservation of ancient Hindu manuscripts

When 13th-century Indian religious scholars painstakingly incised spiritual philosophies on palm fronds to preserve for future generations, they could not have imagined that a team of 21st-century scholars would one day be working with high-performance computers to further preserve and make them accessible. This is what researchers from the Center for Preservation of Ancient Manuscripts (CPAM) at Rochester Institute of Technology (RIT) are doing for ancient Hindu documents, the Sarvamoola Granthas.

The Sarvamoola Granthas contain the philosophical teachings of Shri Madvacharya, a proponent of the dualist Hindu philosophy, Dvaita. Madvacharya is among the most influential teachers in Hindu religious history, credited with ushering in a new spiritual age in India.

For centuries, the palm-leaf documents of the Sarvamoola Granthas here housed in Indian monasteries called Mathas. With Palm Fronds exposure to damaging atmospheric elements, over time the leaves became brittle, discolored and difficult to read. Scholars were in danger of losing access to their content. Today, the CPAM team is working to capture the document images in a digital format (see above photo), to better endure the passage of time and to once again be made readable.

“The Sarvamoola Granthas project is unique in that it uses some of the most advanced modern technology to preserve irreplaceable ancient knowledge,” says CPAM director P.R. Mukund, who leads the project along with co-director Roger Easton, distinguished researcher Ajay Pasupuleti and post-doctoral researcher Sri Priya Das. “Without digital preservation, it is very likely that more of these important documents would be lost to future scholars.”

The process of digitizing so many detailed images was no minor task, producing more than two terabytes of data, a large amount of data for historical research, and more than local facilities at RIT could easily manage. CPAM needed a quick, reliable way to move these precious data-sets to where they could be stored safely and replicated easily. They turned to TeraGrid’s Data Capacitor, a system developed by IU to temporarily store, transfer and manipulate very large data sets. Mukund worked closely with Stephen Simms, IU TeraGrid site lead and manager of the Data Capacitor project, to develop a data-management strategy.

“The ability to transfer high-quality images was a daunting task that was made easy thanks to the TeraGrid and the Data Capacitor,” says Mukund. “It was the combination of having access to this state-of-the-art technology, along with the personal support we received, that allowed us to successfully manage the data.”

Using the Lustre-WAN implementation (developed by IU in collaboration with DDN, Inc), data were transferred from RIT through the New York State Education and Research Network at an achieved rate between 80 and 110 megabits per second, roughly 80-percent of the peak theoretical throughput possible on the one-gigabit connection. Once copied to the Data Capacitor, the data were backed up to two tape silos in different locations within IU’s Massive Data Storage System, to ensure the data would remain secure.

The Sarvamoola Granthas project was one of several projects highlighted by the winning team at the Bandwidth Challenge, an annual competition testing the limits of modern supercomputers, held during the Supercomputing ’07 conference. Using the Data Capacitor, a team from IU, PSC and ORNL claimed first place.

“While most of the projects we support are in the sciences,” says Simms, “it was exciting to demonstrate how this type of technology also has very real and important applications in the humanities. This project showed the general capability of Lustre-WAN clients implemented by IU to move data rapidly and easily from an instrument outside of the TeraGrid into the TeraGrid.”

Looking for light: TeraGrid resources enable new understanding of Alzheimer’s disease

Structure of the amyloid-ß protein

As many as 5.2-million Americans live with Alzheimer’s disease—according to a 2008 report by the Alzheimer’s Association. World-wide it is the sixth leading cause of death. Commonly described as “going dark,” Alzheimer’s can make it difficult or impossible to live a fully independent life.

Using technology made available on the TeraGrid, Mu-Hyun Baik, associate professor of chemistry and informatics at Indiana University (IU), is working on the front lines of Alzheimer’s research, helping to bring light to Alzheimer’s patients and their loved ones.

Currently, nobody knows with certainty what microscopic events lead to brain damage. With support from the Research Corporation and Alfred E. Sloan Foundation, however, Baik’s work has simulated the structure of the amyloid-ß protein, widely believed to be the cause of Alzheimer’s disease.

“At the moment, scientists are fishing in the dark,” says Baik. “We don’t even know what the amyloid-ß deposits in the brain look like. Rationally thinking about a treatment or even a cure is impossible. The only hope in this situation is that we find a treatment by accident, which is a very shaky proposition. We need to better understand how amyloid-ß behaves chemically in the brain in order to find a systematic path to treatment and cure.”

This type of fundamental research requires enormous computational infrastructure. “The combination of the advanced computing capability of the TeraGrid,” says Baik, “and the high performance data-storage solution provided by the Data Capacitor was critical. Without it, I would have never dared to begin this project.”

Working with TeraGrid staff, Baik’s amyloid-ß protein analysis was achieved with a workflow that involved using the Data Capacitor, a TeraGrid resource developed at IU, along with IU’s Big Red supercomputer. The massive volumes of data produced by Baik’s computational experiments were managed using a Lustre wide-area file system—an approach that was largely untested at the time.

The payoff for this experimental approach was huge, as Baik demonstrated along with a team from IU, PSC, and ORNL at the Bandwidth Challenge competition held at the 2007 Supercomputing conference in Reno, Nevada. Running data from Baik’s simulations along with a variety of other workflows, the Data Capacitor achieved a staggering bi-directional data transfer rate of 18.2 gigabits per second out of a possible 20. The team clinched the competition using this new model for gathering data from remote resources, and demonstrated the potential of TeraGrid resources for supporting this type of data intensive research.

“The ability to move so much data quickly and easily has already been a tremendous benefit to our research,” says Baik. “We have recently completed the first phase of our work and proposed for the very first time a high-resolution structure of the amyloid-ß deposit. This work is currently being reviewed for publication and our initial conclusions are quite unexpected. We’re on a path toward making some very important discoveries that could change the lives of those suffering from Alzheimer’s disease. This was a new direction in my research that was only possible because I have access to TeraGrid resources.”

More information:
http://info.chem.indiana.edu/sb/page/normal/757.html

New Light

Mu-Hyun “Mookie” Baik

The cause of Alzheimer’s disease is not yet identified and there is currently neither an understanding of the pathogenesis of the disease nor a rational strategy towards a treatment. It is clear that deposits in the brain consisting of small peptides identified as beta-amyloids play some role in the destruction of brain cells that ultimately leads to dementia and complete loss of brain function. One of the leading hypothesis is that the beta-amyloid deposits bind copper, which then catalyzes the generation of hydrogen peroxide from dioxygen, which causes oxidative damage. Unfortunately, the structure of these Cu-amyloid deposits are not available and nobody knows how copper binds to these peptides.

IU Professor, Mu-Hyun “Mookie” Baik and coworkers have used TeraGrid resources, the NSF-funded Data Capacitor and the IU’s Big Red supercomputer to construct a large scale computer model of these amyloid entities and examine the copper binding behavior. For the first time, a high-resolution structure of the copper bound beta-amyloid was obtained and communicated in a recent publication in the Journal of Biological Inorganic Chemistry in fall of 2008.

In addition to providing structural information, Baik found that the copper binding behavior is highly dependent on the length of the peptide oligomer. The biologically relevant amyloid is a 42-residue paptide, which is not very soluble. For experimental convenience many researchers use a close analogue, a 40-residue peptide, as a model to the biologically important amyloid. Baik showed for the first time that the copper bound structures of these two analogous peptides are dramatically different, providing a possible explanation for the dramatically different results and controversies reported in the literature, where peptides of different lengths were used as models. Baik’s structural work provides a foundation for future work within Baik’s and other research groups to address structural and reactivity inquiries. Baik’s recently released results will fundamentally change the way medical researchers further study Alzheimer’s disease and move toward the development of more effective treatments. Baik’s potentially life-saving work could not have been accomplished without access to NSF-funded technology resources.