Dell Launches HPC Infrastructure to Power Genomic Research
Dell has launched Active Infrastructure for High-Performance Computing (HPC) Life Sciences, a platform that allows medical researchers to collect and analyze genomic data.
Clustered blade servers and a high-performance file system will allow researchers to process a complete genomic analysis in less than one day, according to Dell, which announced the product at its Enterprise Forum in San Jose, Calif., on June 4.
The HPC Life Sciences version of Active Infrastructure is an open-standards architecture that consists of a Linux-based 512-node computing cluster and a Lustre parallel-processing system. The infrastructure is built into a single 42-unit rack with 1.5 terabytes of memory.
Along with several other products, on June 4 Dell also introduced its PowerEdge VRTX private cloud-in-a-box platform for small businesses. The PowerEdge platform is suitable for small health care practices, rural hospital satellite offices and offsite health clinics, according to Dell. VRTX combines servers, storage, networking and management hardware in a single chassis.
"You can imagine this device being the virtualization server to manage desktops in a local clinic," Dr. Andrew Litt, Dell's chief medical officer, told eWEEK. Clinics could use the platform to store, process, manage and share radiology images.
Active Infrastructure for HPC Life Sciences will be available in August, and PowerEdge will be available June 26.
Technology such as Active Infrastructure could make genome sequencing faster and more accessible, according to Litt. By making the collection of genomic data less costly, the infrastructure can help the sequencing of complete genomes become more common. "Genomic sequencing is a big deal, and it's rapidly spreading," Litt said.
"We've moved from the period in the last five years or so where people have been looking at specific genes or specific genetic variations like the BRCA gene to do entire genomic sequencing and trying to make evaluations from that."
Detecting the BRCA gene enables doctors to predict women's inherited risk of developing breast and ovarian cancers.
A single genome comprises 100GB to 200GB of data, Dell reported.
The HPC infrastructure can process up to 38 genomes per day and 266 genomes per week, Dell reported.
Optimizing an integrated HPC architecture rather than off-the-shelf components makes the speed of genomic sequencing possible, according to Litt. Packaging hardware in an HPC infrastructure also leads to more efficient troubleshooting, he noted.
"From an efficiency point of view, you plug it in and you're up and running," Litt said. "It all works in one package, so you don't have to call three people when it doesn't work."
HPC will be important to health care IT as the industry grows more complex, according to Litt.
"If you've got an IT department, you can think about how to add value for the researchers and institutions and the physicians rather than worrying about the blinking lights and maintaining the servers," he said.
Dell has been working with the Translational Genomics Research Institute (TGen) to reduce genomic analytics time for pediatric cancer clinical trials. HPC has resulted in a "twelve-fold" improvement in the ability to process patient data, Dell reported.
Although the cost to do genomic sequencing is currently in the thousands for each genome, this price should come down as the process gets more common, Litt predicted.
Academic institutions and medical research labs face the challenge of building the infrastructure that can handle the computing and analysis necessary for genomic data. "It's not like buying a PC and turning it on," Litt said. "It's pretty complicated—you need a well-synced architecture where everything is running together."
Without an integrated genomic architecture, sequencing can take several days instead of a few hours, according to Dell.
If medical facilities adopt an integrated infrastructure that can handle the large amounts of genomic data, every cancer patient that enters facilities could be sequenced to gain data that can boost clinical care, Litt said.