Celera Genomics’ bioinformatics subsidiary Paracel quietly threw in the towel last week, posting a notice on its website that it had ceased operations — except for ongoing support for its current customers — as of Sept. 28.
The closure brought an end to several years of uncertainty about the business unit’s role within Celera after the genomics giant reinvented itself as a drug discovery shop in mid-2002.
“The decision to close Paracel was made following an analysis of various alternatives for the business,” Celera noted in a statement on its website.
“Applera purchased Paracel in June 2000 because its advanced genomic supercomputing technology showed promise in applications such as Celera Genomics’ effort to assemble the human genome, and for third-party genomic and text-finding applications,” the statement continued. “As a result of the shift in focus at Celera from genomics data and services to the development of targeted therapeutics, Paracel is no longer deemed strategic to Celera’s business plan.”
Celera said that it would provide further details of the closure in its next quarterly earnings statement.
Amanda McNulty Sheldon, director of marketing at Paracel, said that the company “pursued a variety of different options, including a sale, before the decision was made, but this is their final decision.”
Paracel’s intellectual property, including several patents, is owned by Celera, Sheldon said, adding that “interested parties can contact Paracel.”
Paracel currently employs around 40 people, including contractors, Sheldon said. This team will be cut to a “core staff” of around eight people who will be responsible for supporting the company’s existing customer base. Sheldon said that Paracel has around 100 life science contracts in place currently, and that the “majority” of them will end within the year.
“Our primary focus will be to continue to offer support to our customers and make this as easy as a transition as possible,” Sheldon said. As for the specifics of that transition, “each individual customer will have their own needs,” she said. “As there are partners that we can work with to ease the transition, we’ll be doing that, or we will provide the opportunity to obtain the support they need to transition it in-house.”
Living Up to Expectations
It’s questionable whether Paracel would have ever been able to generate enough revenue to justify Celera’s purchase of the firm for approximately 2.26 million shares of stock in 2000, worth in the neighborhood of $280 million at the time.
Within a year after the acquisition, Celera was already blaming the bioinformatics firm for a good part of its net losses, and began writing off the value of Paracel through the impairment of goodwill and other intangible assets [BioInform 08-06-01].
Celera said that charges related to the closure of the business and the completion of remaining service obligations would occur over the next several quarters.
Paracel was founded in 1992 to commercialize technology called Fast Data Filtering developed by TRW for government applications. Together with researchers from Caltech and Applied Biosystems, the firm programmed the FDF system for genomic data analysis.
The company’s flagship bioinformatics product, GeneMatcher, led a field of several “accelerated” bioinformatics hardware systems that used reconfigurable chips to speed compute-intensive algorithms like Smith-Waterman and hidden Markov models.
More recently, as demand softened for these dedicated systems as part of the decline in the broader bioinformatics market, Paracel shifted its strategy toward the emerging Linux cluster trend. It began selling Blast and other bioinformatics algorithms pre-installed on its Cyclone Linux cluster, and then late last year began offering the clusters as standalone general-purpose systems [BioInform 09-01-03]. The company also attempted to round out its product offering — and target the broader market of end-user biologists — with a pathway visualization software package called PathWorks that it launched in mid-2003 [BioInform 05-19-03].
But these moves proved to be too little too late for the firm, which just announced in July that it had signed three new customers in Europe for its GeneMatcher2 sequence analysis system — the Nestlé Research Center in Switzerland, the University of Wuerzburg in Germany, and the Institute of Molecular Pathology in Austria.
Despite these recent customer wins, industry observers said that the company’s closure came as no surprise. “They’ve been very low key in the last few years,” said Simchon Faigler, CEO of Biocceleration, a bioinformatics computing firm that Compugen divested last year.
Faigler said that the market is indeed shifting away from hardware-based accelerated systems such as GeneMatcher and even his company’s own BioXL/H system. “Specific solutions are going to disappear in a few years, and the next big thing will be solutions based on general-purpose computing systems,” he said.
Biocceleration plans to introduce a series of new bioinformatics products built on commodity components, Faigler said, but “unlike Paracel — which was also trying to compete with the Linux clusters themselves — we are not going to compete on Linux clusters because there are very large and significant players there, like Dell, IBM, HP, and others. We are only going to focus on the applications — on the software that enables [users] to turn these Linux clusters into accelerators of database searches.”
Paracel’s Sheldon agreed that the Linux cluster market proved to be a bit more competitive than the company had bargained for. “There are obviously a lot more players in the general computing market,” she said.
Other vendors of reconfigurable chip-based hardware systems said that there is still plenty of potential for this technology, although perhaps not yet in bioinformatics. Ed McGarr, vice president of sales and customer services at Starbridge Systems, which began selling its own FPGA-based platform to life science customers earlier this year [BioInform 04-12-04], said that the company’s Hypercomputer is selling well in a number of more “mature” industries like signal processing and materials science, but “there’s a slower pick-up in the life sciences,” he said.
“The technology is applicable across industries, so from our perspective, we go where the interest is,” McGarr said. “I think there is initial early interest in the bio-IT market, but it’s not quite evolved yet.”