The high-performance computer cluster, which was donated to the Phoenix, Ariz.-based Translational Genomics Research Institute (TGen), will allow researchers to quickly access more precise data on diagnostic images, genetic sequencing of the RNA in tumors, and genetic variations that contribute to the disease.
Dell began building the cluster in December using its PowerEdge M420 Blade Server, which supports a parallel approach to computation. With this technology, TGen can comprehensively analyze a patient's tumor RNA profile in one day, compared to the seven days that were previously required--an important advantage in battling aggressive childhood cancers like neuroblastoma.
Dell officials also said the company will begin to connect the biomedical researchers who are sequencing and analyzing tumors at TGen with oncologists treating patients participating in the trial at 14 medical centers. Through the use of cloud technology, Dell's computer cluster will further aid researchers by eliminating the need to express mail hard drives containing tumor and diagnostic images and genomic sequencing data between locations.
[ Which medical apps are doctors and patients turning to? See 9 Mobile Health Apps Worth A Closer Look. ]
According to Jamie Coffin, vice president and general manager of Dell Healthcare and Life Sciences, biomedical researchers at TGen will compare the RNA sequencing profiles of healthy cells against tumor cells to identify biomarkers and potential tumor vulnerabilities.
Physicians will then check those vulnerabilities against an index of drugs approved for cancer trials to identify the most effective treatment based on the genetic makeup of the child's tumor. Coffin said physicians will also input treatment outcomes into the cloud so that over time, they will have a body of knowledge on recurring biomarkers and the most effective treatments.
TGen has developed algorithms that will comb through drug indexes from approved cancer clinical trials, genetic sequencing of tumor and healthy cells for comparison, treatment findings from the Van Andel Research Institute, and tumor samples and diagnostic patient information from the 14 participating medical centers.
To demonstrate the amount of data and computing power needed to do such sophisticated research, Coffin said to sequence one child's RNA creates about 100 million reads per sample. (A read is a string of 50 or so "bases" or genetic building blocks.) In the case of RNA sequencing, the number of reads per gene gives some indication of how much of a particular gene is expressed.
The estimated size of the computation and collaboration cloud necessary to support the clinical trial for pediatric cancer is approximately 9 teraflops, which will grow over time to approximately 13 teraflops as the program adds clinical sites and more children are enrolled in the trial, Coffin said.
As the data increases, so too does the need for data storage capacity.
"We will increase our storage from 300 terabytes today to 1 petabyte by the end of year. Currently a handful of children are participating in the trial," Coffin said in an interview with InformationWeek Healthcare. "Additional processing capacity will support participation expansion to 100 children by the end of next year and 600 children in year 3 of the program."
In addition to the biomedical researchers from TGen, oncologists from the Neuroblastoma and Medulloblastoma Translational Research Consortium (NMTRC), based in Grand Rapids, Mich., will also have access to the new high-performance computing cluster.
During the last few decades, approved treatments for pediatric cancer have lagged behind those for adults, with only one new treatment for pediatric cancer approved by the FDA since the 1980s, compared with 50 treatments approved for adult cancer in the same timeframe. Coffin hopes that Dell's high-performance computing cluster can speed up the research process.
In the meantime, he identified the cloud-based cluster donated to TGen as a model that can be used to manage large amounts of data in a variety of areas in healthcare.
"Initially piloted in pediatric oncology, where there is a real need and unmet gap, cloud-computing platforms like this one could also support personalized medicine in adult oncology and other disease areas," Coffin said. "We believe that cloud-computing platforms could support complex biomedical knowledge exchanges between healthcare providers, research centers, clinical genomics and molecular diagnostic vendors, and pharmaceutical researchers interested in participating in personalized medicine."
Get the new, all-digital Healthcare CIO 25 issue of InformationWeek Healthcare. It's our second annual honor roll of the health IT leaders driving healthcare's transformation. (Free registration required.)