Editor’s note: Please join a Tweet chat featuring Dr. Murali Ramanathan and other healthcare and data analytics experts May 10 from noon to 1 p.m. Eastern Time at #IBMdatachat.
Multiple sclerosis is a cruel disease. It typically strikes young adults. The body’s own immune system attacks the brain and spinal cord, resulting in physical disabilities, cognitive problems and a host of other chronic symptoms. The cause isn’t known. There is no cure.
Fortunately, the amount of biomedical and clinical data related to MS has exploded over the past decade, and, at the same time, new research methods make it possible to assess environmental factors and hundreds of thousands of genetic variations taken from single samples.
Researchers at The State University of New York at Buffalo are using a new approach to computing in an attempt to identify the causes and promising therapies. “The eventual goal is to help develop a cure or prevention for MS,” says Dr. Murali Ramanathan, a professor of pharmaceutical sciences and neurology at SUNY Buffalo. “The ability to do this kind of computational analysis is a great complement to basic science and clinical research.”
Ramanathan is one of two researchers leading the Data Intensive Discovery Initiative, or DI2, which is focused on developing new algorithms and modeling techniques for analyzing genetic and environmental factors in multiple sclerosis. They also train graduate students.
He and his colleagues believe that MS is caused by a complex combination of gene-to-gene and gene-to-environment interactions. The chief environmental factors include living far from the equator, viral infections and cigarette smoking. They evaluate a wide range of factors, including gender, geography, ethnicity, diet, exercise, sun exposure and living and working conditions. The clinical data include medical records, lab results, MRI scans and patient surveys.
The researchers are using a computer cluster from IBM’s Netezza that combines processing, a database, storage and analytics into a single system, or appliance. This data-intensive architecture makes it possible to handle large amounts of data and derive insights quickly. The Buffalo researchers have found that analyses that once took days can now be completed in mere minutes.
Previously, Ramanathan used conventional parallel processing on a cluster of computer servers to crunch his data. Those systems divide large problems into smaller ones and solve them concurrently using hundreds or thousands of processors. The Netezza system also takes advantage of parallel processing. But, in addition, it uses specialized processor chips to filter the data that’s sitting in on storage disks before passing along only the relevant pieces to the main processors. At the same time, the system performs some of the analysis as the data is moving off the disks, rather than handling all of it on the main processors. So the design, essentially, moves a lot of the processing to the data rather than moving so much data to the microprocessors. As a result, the work can be done faster and more efficiently.
It took Ramanathan months to warm up to this approach to number crunching, but now he’s a huge fan. “I believe that the vast majority of problems in clinical and biomedical research will be addressed through this new way of computing,” he says.