Last month’s Beyond Genome conference was Carlsbad, Calif.-based Omnigon Technology’s “coming out party,” according to CEO Bob McKuin, but it was a very discreet affair: Only a select number of attendees were privy to an invite-only demo of the company’s technology.
The rest of us may have to keep wondering until granted our own private audience with Omnigon’s data analysis system, which, although under development for over two years, hasn’t even been given a name yet. Omnigon claims that its pattern recognition technology, based on an approach it calls Transfinite Representation Theory, is so different from existing approaches that it literally has to be seen to be believed — especially by its target user base of biologists and bioinformaticists.
“Scientists are pretty skeptical. They think there’s nothing new in the world,” said McKuin. “But when they saw what we could do they were absolutely captivated.”
At Beyond Genome, Omnigon demonstrated how its technology could be applied to overlap detection, the first of several bioinformatics applications the company sees for the system. According to Omnigon, the approach is able to eliminate the heuristics and algorithmic acrobatics that are a necessary part of current bioinformatics solutions.
The company describes its technology as a “holistic” approach that is able to find and store data relationships in high-dimensional space. But it’s not just another clustering tool, McKuin noted. A by-product of the process is 300-to-1 data compaction, for example — welcome news for organizations seeking relief from ever-mounting storage requirements.
Company literature illustrates how an algorithm-based process such as Celera’s five-step whole-genome assembly pipeline could be reduced to a single step with Omnigon’s approach. According to the company, Celera’s first screening step, as well as its unitigger, scaffolder, and repeat resolver steps, would be unnecessary because the Omnigon system organizes information according to its relationships prior to analysis. The company has not applied its approach to whole-genome assembly, however, and pointed out that the example is only provided as an illustration of the system’s potential.
“We don’t claim to be domain experts,” said McKuin. Omnigon describes itself as an informatics company, not a bioinformatics company, and has targeted the financial markets, oil and gas, and other compute-heavy industries as future markets for its tools. It has set its sights on the life sciences first, however, because it sits in the “sweet spot” of the system’s data-crunching capabilities, according to McKuin. In addition, the company’s Southern California location places it in the heart of biotech country, making those sales calls and on-site demos that much easier.
McKuin said the patent-pending technology has been verified by several well-known experts in biology, including Pavel Pevzner of the University of California, San Diego, who is advising the company on bioinformatics applications. Pevzner was on vacation and unavailable to comment last week.
Omnigon is in discussions with a number of companies now, McKuin said — all leads from the private showings at Beyond Genome. It intends to work closely with its customers to identify the best ways its technology can be put to use in solving their data analysis problems.