Seeking to raise awareness about its computer appliance in the life sciences and other industries, YarcData, a spinoff from supercomputing firm Cray, launched a challenge last June that offered contestants a chance to use uRiKA, its graph analytics appliance, in their research projects.
The company selected six research groups to have access to their system and to compete for cash prizes totaling $100,000. Earlier this month, they announced that the Institute for Systems Biology won first place and a $70,000 cash prize for using the system to research opportunities to repurpose existing cancer drugs.
Second place and a $13,000 cash prize was awarded to researchers at the University of California, Santa Barbara for a project investigating the causes of autism; and third place and $8,000 went to a team at the University of Washington that used uRiKA to predict cardiovascular patients' risk and mortality in the month following a heart attack. The remaining three finalists who worked on projects in crime, sports performance, and social network analysis received $3,000 prizes for their entries.
Pleasanton, Calif.-based YarcData opened its doors officially last February. In an interview with BioInform this week, company representatives said that the company spun out from Cray because it offers a product and operates a business model that’s different from its parent.
Misti Lusher, YarcData's director of marketing, explained that unlike Cray, which markets large, costly supercomputers to large research institutions and government entities, YarcData offers a much smaller, cheaper, and more user-friendly enterprise system that doesn’t require as much IT expertise and fits easily into customers' data centers.
Furthermore, uRiKA is aimed toward customers that want a solution that lets them solve specific problems rather than a general high-performance computing solution, explained David Anstey, YarcData's global head of life sciences.
uRiKA provides between 2 terabytes and 512 terabytes of shared memory, multithreaded graph processors supporting 128 threads per processor, and a semantic database that represents data as triples and makes it easier to see associations between data. The system supports ad hoc queries, pattern-based searches, and inferencing, and it also accepts and incorporates new data into existing graphs. It complements existing data warehouse or Hadoop clusters? by offloading graph workloads and interoperating within a customers' existing analytics workflow.
YarcData offers two purchasing models for uRiKA. Customers can either buy a system or they can pay for an annual or multi-year subscription that will offer on-premise access to the system for the duration of the subscription.
The latter option is likely to appeal to small and mid-sized pharmaceutical companies who may not have the capital to pay full price for a system, as well as academic centers whose expenditure is determined by grant funding. It could also appeal to organizations looking to test the utility of the system before making a commitment, the company said.
YarcData did not disclose specific price points for either option although Anstey did tell BioInform that the pricing will vary depending on the size of the machine and customers specifications.
Although it's only just dipped its toe into the space, YarcData believes that sales to life sciences-based institutions and companies will ultimately make up a large portion of its business where customers are seeking compute power combine and analyze large datasets from multiple sources.
It will have to compete with hardware vendors such as SAP and Oracle who also offer compute appliances that they say can help researchers analyze large datasets.
However, YarcData believes that its graph-based approach to analysis can compete well with these solutions — which use different underlying approaches — because it provides users with a graphical representation of the connections in their data making it easier to see unknown associations. They can also update these graphs with new data as it becomes available, all of which helps researchers generate and test or revise their hypothesis faster.
It also says that uRiKA is much cheaper than other big data computing solutions which should make it more attractive for budget-conscious customers.
The company also believes it can outcompete hardware firms that offer graph-based analytics appliances because its solution is much more scalable.
"The problem with graph analytics is that there has never really been technology that could scale" to handle very large amounts of data, Lusher said. "We built this appliance purposely to deal with graphs in a very scalable fashion."
Also, YarcData began developing its solution about four years ago which gives it at least an 18-month lead over other appliance vendors that are looking into the space, she said.
In addition to life sciences, YarcData has customers in financial services and government. Customers include the Canadian and United States governments, Mayo Clinic, Noblis, Pittsburgh Supercomputing Center, and Sandia National Laboratories.