Big Win for Grid Computing as Celera And Parabon, Accelrys and UD Pair Up

With two recent deals — one between Parabon Computation and Celera Genomics and another between United Devices and Accelrys — it appears that distributed computing technology has transitioned from an experimental curiosity to an accepted high-performance computing platform for life sciences research.

While pharmaceutical and biotech companies have been slow to adopt distributed or “grid” platforms, which harness the unused compute cycles of networked desktop PCs, proponents see these key agreements as validation of the technology’s ability to compete with more established high-performance computing approaches.

Austin, Texas-based United Devices, which refers to its technology as “edge distributed” computing to distinguish it from clusters and compute farms, is partnering with Accelrys to combine Accelrys’ life science software applications with UD’s MetaProcessor distributed platform. The companies will first port Accelrys’ CHARMm macromolecular mechanics and dynamics software and its LigandFit software for computer-aided drug design and docking.

Meanwhile, Parabon of Fairfax, Va., will partner with Celera to jointly develop proteomics applications to run on Parabon’s Frontier platform. The first product will be an application to identify proteins from tandem mass spectrometry data.

Financial terms were not disclosed for either agreement.

WEB-BASED Cancer Project SERVED as Pilot

Scott Kahn, senior vice president of life sciences at Accelrys, said the decision to partner with UD was based on customer demand for applications that could run on a grid. “Typically we let the customers drive the platform decision, and this is one that we’ve been hearing from our customers that’s of interest to them,” he said.

Kahn noted that Accelrys evaluated the technology’s potential by observing the online cancer research project UD is supporting in partnership with Intel and the University of Oxford. Over a million PCs across the Web are running the project’s THINK screensaver program to screen small molecules against protein targets. “That kind of application is indicative of the type of code we have,” Kahn said. “That more than adequately satisfied us in terms of its reliability and robustness.”

Kahn said he expects application performance to scale linearly with the number of processors.

UD and Accelrys will co-market each other’s products and services as part of the agreement.

David Wilson, vice president of marketing and business development at UD, described the Accelrys partnership as “huge validation that they recognize our leadership in this space and they believe in the benefits of edge distributed computing technology for their customer base.”

He said that enlisting the support of an independent software vendor is a key part of the company’s strategy to cater to the life sciences market. UD has published a software developer’s kit that Wilson said could “easily” port public or in-house applications. The company used the kit to optimize Blast and Hmmer for the MetaProcessor platform and plans to add to its list of publicly available software.

But the deal with Accelrys now gives UD customers the opportunity to run third-party applications on the platform as well. LigandFit, in particular, should help the company gain entry into the pharmaceutical sector because it addresses virtual screening — the computationally demanding “sweet spot” of the life sciences market that Wilson said UD is targeting with its platform.

Parabon Powers Proteomics  

Celera, which has already installed a massive Compaq compute farm for its work in genomics, opted for Parabon’s solution to add a bit of a boost to its proteomics efforts.

The deal makes Celera the first life sciences company to move beyond the pilot phase and adopt the technology as part of its compute infrastructure.

John Reynders, Celera’s vice president of information systems, called Frontier “an ideal complement to the industrial scale, compute intensive, bioinformatics platform” that the company has in place.

Over the Hump?

While the value proposition of accessing compute power that already exists on the desktops of an organization may seem evident, grid computing has not yet caught on within the industry.

UD’s Wilson said that life science clients are taking a “walk, don’t run” approach for several reasons. Issues of safety and unobtrusiveness are key, he said, as are considerations of how well the platform is able to deliver scientifically validated results. Most potential clients run pilot studies to work out these issues that can take several months, which Wilson deemed “a natural enterprise sales cycle.”

Some critics of the technology point out that it is effective for only a narrow range of computing tasks, but Wilson was quick to admit that while some applications thrive in the environment, “others are better suited to running on a cluster or a supercomputer in a data center.”

UD has pegged approximately one-third of the current high-performance computing market — $298 million out of a total $902 million for life sciences — as the addressable market for its approach.

Avaki Enters the Game

As further evidence that the market potential within this area is considerable, Avaki, a grid computing company based in Cambridge, Mass., announced its intention to target the life sciences computing market with its products last week. The company recently closed on $16 million in funding and launched Avaki 2.0, software that adds data sharing capabilities to the distributed computing infrastructure.

Dave Fish, Avaki CEO, said the company is “proactively focusing on the life sciences as our primary market” in the coming year.

In addition, Entropia, another distributed computing vendor, recently entered a partnership with TurboGenomics to enable the company’s TurboBlast sequence comparison software to run on its platform. Entropia said further partnerships with other third-party software vendors are planned.

But the added competition doesn’t have UD’s Wilson worried. “We certainly have a lead on them,” he said. “If we just continue to make progress with the traction that we have, we’ll make if awfully tough for somebody like Avaki to penetrate the market.”

And as far as competition from Entropia, Wilson cited UD’s success with the web-based cancer project as a key differentiator. “We''ve already scaled this thing bigger than any enterprise customer is going to scale it to,” he said. “We''ve proven the scalability and reliability of it across the Internet — the most diverse, hostile network that you can imagine.”

— BT