TORONTO — The computing demands of the life sciences industry are pushing drug discovery companies to explore grid computing as a means of leveraging their IT investments, executives at the BIO 2002 conference said Tuesday.
put on this week at the Metro Toronto Convention Centre by the Biotechnology Industry Organization (BIO), the conference features workshops on drug development, food and agriculture, regulatory processes and bioethics. Perhaps more than at any time in the conference’s 16-year history, however, data management and storage have emerged as key challenges facing startups and established firms.
“”It was only 10 years ago we started speaking (about) computers in life sciences,”” said Howard Asher, director of the global life sciences practice at Sun Microsystems Inc. “”No one really forecast what would happen in terms of the data that these companies would produce.””
The drug discovery companies are focused on a range of research areas, but some of the most recent startups have concentrated their efforts on proteomics — the study of how proteins function within the gene. Biotechnology experts say this field, which could provide a much deeper understanding of how diseases are caused and can be treated, is at least twice as complex as genomics, the study and mapping of the human genome.
Mike Swenson, an analyst with IDC, said the supercomputing required to process the massive amounts of data involved has triggered “”bio grids,”” that build upon the grid computing approaches of the public sector. Grid computing allows data, power and applications to be shared across organizations and continents. Government-run organizations and research labs in universities have emerged as the early adopters of this technique, as it helps them pool IT resources for compute-intensive projects.
Security concerns and fear of leaking competitive information has kept the private sector from forming cross-company grids, Swenson said. “”There is a resistance to using external resources. With the drug discovery companies, their value is intellectual property,”” he said. “”Even if you can get your IT department to say, yes, this is a good idea, senior management may say no.””
Self-contained grids, on the other hand, are developing behind secure firewalls in some firms, Sorensen said. Life science companies are often the ideal candidates for this approach because they typically own many desktops, workstations and servers with several CPUs that aren’t used for large parts of the day. These enterprises are atypical because many of them have been built from scratch with state-of-the-art equipment; there is no legacy infrastructure.
Lloyd Segal, CEO of proteomics specialist Caprion Pharmaceuticals Inc., described himself as “”the poster child”” of grid computing in life sciences. Caprion has about 80 CPUs running on its systems throughout its facility in Dorval, Que. Many of its staff work at night, which means the combined computing power of those CPUs are available for much of the day, Segal said.
“”Grid computing is like the teenage sex of proteomics: everyone’s talking about it, but no one’s doing it,”” Segal said. “”I could take that analogy further, but I’d probably get into trouble.””
Tom Kassberg, CEO of Plexxicon in Berkeley, Calif., said his firm fuses proteomics with chemistry, searching for the microwave compounds used in building drugs. The company has about 60 CPUs in its small office. Grid computing (based, like Caprion, upon Sun’s servers and grid software applications) gives Plexxicon both scale and flexibility, he said.
“”We’re working on four projects today, but grid computing allows us to prioritize,”” he said. “”We can give some scientists greater access (to computing power) over others.””
Plexxicon employs 40 people now, but is preparing to expand to 120 people at least, Kassberg added. Grid computing helps these companies prepare for their growth curve, according to Swenson, because it stretches the longevity of the IT asset. “”It really captures the value of the existing server infrastructure,”” he said. “”For some companies, it may mean being able to put off buying another server for a year.””
Segal said grids make growth more manageable because each time Caprion hires a new bioinformatician, they receive a desktop with at least two CPUs on it. This adds to the grid. “”It allowed us to draw a path that’s more incremental,”” he said.
Asher described the market as a horizontal landscape of idle computing that can be put to work, potentially accelerating the pace at which diseases are cured. “”They know the crown jewel is out there,”” he said. “”They just have to find it.””