Nine months after creating a life sciences group within its high-performance computing business unit, Intel has kicked off its first large-scale bioinformatics project — a collaboration with the Swiss Institute of Bioinformatics, Hewlett-Packard, and several Swiss universities to create a high-performance computing center dedicated to life science research.
The project, dubbed Vital-IT, is not only an initiative to provide high-performance computing resources for researchers in the Geneva area — a region known as the “Swiss Biotech Valley” — but also a way for Intel and HP to test out the new Itanium processor in the life sciences market. HP will build an Itanium 2-based cluster at the University of Lausanne that will initially be used by researchers at SIB, as well as the Swiss Federal Institute of Technology in Lausanne (EPFL), and the Universities of Lausanne, Geneva, and Basel. Eventually, the partners said they plan to make the Vital-IT resources available to academic and commercial users in the Swiss Biotech Valley and elsewhere.
Intel will pitch in by working with SIB developers to tweak bioinformatics applications for the Itanium 2 processor. This involves porting existing code to the architecture as well as developing new applications that are optimized for the Intel platform.
Vital-IT is initially funded for three years, with a total budget of $1.5 million per year. Intel, HP, and SIB are financially supporting the project, but the bulk of the funding is from the Universities of Geneva and Lausanne and the EPFL, according to Ron Appel, chairman of the SIB executive board and a founding participant in the Vital-IT project.
The participants have not yet determined the size of the cluster to be built, Appel said.
A Vital Need for IT
The SIB, which spawned Swiss-Prot, ExPasy, and other bioinformatics essentials, has its fair share of computational tools and servers, but “each group has its own resources,” Appel told BioInform. “What is really missing is large, centralized resources.”
Appel said that although the SIB and its academic partners already had a lot to offer in terms of bioinformatics, genomics, and proteomics expertise, “we thought it would be a good thing to put together a high-performance computing center for the life sciences that would offer a fairly large machine, and around the machine, a small team of specialists to help the scientists make use of the resources.”
SIB has set up a dedicated research group led by Marie-Christine Sawley and Roberto Fabbretti to manage the project. Appel estimated that around seven SIB researchers would eventually make up the core Vital-IT research team.
Appel said that the project partners evaluated a number of different computational platforms before settling on Intel’s Itanium. “It’s a bet on the future of computing and processors because this is a fairly new technology,” he admitted. “For companies like HP and Intel, the future is Itanium, so we prefer to work on what seems to be the technology of the future.” Within months, Appel predicted, the Itanium “will be the most powerful technology and the most adapted to bioinformatics applications.”
Pushing the Envelope
The Vital-IT project will play a large part in determining whether Appel’s bet on Itanium pays off. The collaboration “will be useful for HP and Intel because there will be heavy users who will push and test these technologies to the limit,” he said.
Rick Hermann, director of Intel’s HPC unit, agreed that the collaboration is a key part of the company’s strategy to secure a larger customer base for the still-unproven Itanium technology in the life sciences HPC market. Noting the overall trend in the recent Top500 supercomputing ranking toward commercial-off-the-shelf-based systems (see story, p. 3), Hermann said, “The expectation is that we’ll start to see a similar type of trend in the life sciences space. Some of the deep collaborations that we do will help accelerate that movement toward standards-based platforms that provide more cost-effective solutions for life sciences researchers.”
The Vital-IT collaboration is similar to high-performance computing projects Intel has carried out in other domain areas. In a collaboration with IBM in Dubai, for example, the company built a solutions center to better understand the needs of end users in the oil and gas industry. Likewise, Intel has partnered with Microsoft and the Cornell Theory Center in New York to build a high-performance computing competency center on Wall Street to work with end-users in the financial computing industry.
Hermann said that projects like this help the chipmaker understand the applications specific to different end-user markets. “We’re able to take that feedback and build it into our thinking relative to future designs around products and architectures and working with our OEMs,” he said. Areas like interconnect technologies, cluster management tools, and methodologies for systems management are most likely to benefit from the collaboration. In one example, he noted, “HPC customers communicated back to us that they would like a variation on the current Itanium product with a different mix of frequency and cache at a price/performance point that provides best-in-class dollar per gigaflop in the industry.” Last week, Hermann said, the company introduced a variation on the Itanium Madison processor that incorporated these changes.
Hermann said that the lake Geneva region is one of several “critical showcase areas around biotechnology,” and that the collaboration with SIB is likely the first of several similar initiatives in the sector.
More than Hardware
While it will provide a new degree of computational power for life science researchers in the SIB community, “the focus of the Vital-IT project is not just the hardware,” Appel stressed. Through the project, SIB will have access not only to the high-performance computing expertise of HP and Intel, but data-mining experts from the University of Geneva and modeling and simulation specialists at EPFL, “so the idea is to put all these competencies together to create new things, new ideas, around high-performance computing,” Appel said.
“It’s not just a question of computer performance, it’s a question of algorithms,” he said. “I believe we can develop new algorithms that we didn’t think about before, just because we didn’t have the resources.” New approaches to handling large volumes of data, separating signal from noise, and extracting pertinent information from huge databases will be top on the priority list, Appel said.
One specific goal of the project is to provide the computational infrastructure for an experimental genomics and proteomics platform that the University of Geneva, the University of Lausanne, and EPFL plan to make available to life scientists in the area, Appel said.
Another goal is to transfer any new technology that is developed to the community “either through an open source environment or through commercial distribution,” Appel said. GeneBio, the commercial agent for SIB that currently distributes SwissProt, may eventually market the technology to commercial users, but Appel said that the tech-transfer details for the project are still under discussion. “It’s not absolutely certain that [GeneBio] will be the one, but it’s a natural possibility,” he said.