May 15, 2007 | With so much data to handle, store, and process, next-generation sequencing offers major opportunities for IT hardware and software companies. Deepak Thakkar, SGI’s BioScience Segment marketing manager, says that SGI provides “hybrid computing with shared memory platforms and industry standard clusters, tied together with specialized computing using technologies like RASC — very suitable for applications like accelerating BLAST, Smith-Waterman.”
SGI’s Bio Solution includes an SMP (symmetric multiprocessing) system, clusters, and optimized RASC coupled with storage. “It’s a complete open solution based on industry standard, Linux, Microsoft PCS, but optimized so we’re able to squeeze maximal price performance from all these systems,” says Thakkar. Pricing starts at under $30,000 for a 16-processor cluster system, with 16 GB memory and almost 4 TB of storage. The solution scales up to include specialized functionalities such as field programmable gate arrays (FPGA’s), SGI RASC, additional storage, shared memory, and cluster compute components.
“We’ve spent great effort in positioning ourselves as the green company,” Thakkar says. It’s “mind boggling that customers are spending up to half their IT budget on power consumption and cooling.” SGI claims significant cooling advantages over racks from competitors, not to mention fourfold space saving.
SGI has installations at UC Riverside, Arizona University, and genome institutes in Malaysia and China, all of which are “gearing up for next-generation systems.” Thakkar notes that another client, the Australian Genome Institute in Melbourne, just purchased a 454 system.
But how are SGI and others going to handle the 1 TB data/day generated by the Illumina and other systems?
Data Management
“We provide software for storage management,” says Thakkar. “There are 3-4 levels of storing data so you can prioritize where data is stored and how frequently you need to access it.” The cluster-based offering stores images, trace files, superimposes data and performs similarity searches.
Illumina recognizes that this is a major issue, says Thakkar: “Unless companies like Illumina provide recommendations and customized solutions for their customers, with partners like SGI, they’re going to see a major bottleneck in terms of data processing.” Left unsaid is the negative impact that would have on consumable spending.
For the prodigious data output by Applied Bio’s SOLiD system, Thakkar sees a “great opportunity” — though Applied Bio is currently working with Dell. Researchers are “appalled at the idea of throwing data away,” he says. “If we can provide a cost-effective storage solution, it’s a case of economics. If for ‘X’ dollars, you can save this information for reference, and satisfy the needs the FDA will impose eventually, then doesn’t it make sense? We might reach a point where we can’t save everything, but we can save 50%.”
Thakkar thinks SGI’s focus on life sciences gives an advantage over companies like IBM and HP, providing specialist solutions and lower price points. “I daresay SGI is taking a rather revolutionary path.” There are several vendors with which SGI could collaborate, ticking off DNAStar, InforSense, Teranode, and Canada’s eXludus. “Their software distributes particular jobs, e.g. alignment, search, in the cluster in a very selective manner so you can utilize the maximum CPU power of the clusters. It can increase productivity in some cases by 4-5 fold.” SGI is also working with Mitrionics, which provides software solutions to accelerate BLAST. With the SGI RASC, “BLAST jobs can be accelerated 20-40X.” -- K.D.
Return to main article.