Bioinformatics a proving ground for IBM: Jasinski

New techniques for generating large amounts of biology-related data have given IT vendors new momentum in penetrating this traditionally low-tech segment of scientific research, a key IBM researcher told attendees at AusBiotech 2002 in Melbourne last week.

Two years after IBM launched a dedicated life sciences division, the company has secured a strong footprint in the emerging market for bioinformatics, according to Dr Joseph Jasinski, senior manager of IBM Research's Computational Biology Centre. Given that the sector has until recently relied on conventional data gathering instead of highly detailed instrumentation, he believes the application of IT analysis techniques will mean a bright future for everyone working in the space.

"I think this field will develop differently than the way computation has developed in other fields of science, since it has developed in a much later time in the computer era," Jasinski explains. "Physicists and chemists started using computers early on, but biologists have not been very compute-intensive because the data one might use to develop computational models hasn't existed until the last decade or so. Getting large amounts of biological data is inherently harder than getting large amounts of other kinds of data."

Jasinski sees bioinformatics as a natural complement to IBM's expertise in data management, since the company has been instrumental in the creation of some of the largest and most powerful applied-science systems in the world. One such system, Blue Gene, laid the foundation for a $US100 million IBM research initiative to tackle the immensely complex challenge of modelling protein folding using a supercomputer that will eventually process over 200 trillion operations per second

That's more computing power than the 500 largest supercomputers in the world today. But raw computing power is only one part of the equation, however. Layered on top of the systems, warns Jasinski, must be the intelligence to appropriately aggregate, analyse and report on the data fed into the system.

"In life sciences, the major issue is not collection but handling data," he explains. "You need to get access to diverse, heterogeneous data sources and allow them to ask questions of those data in an intelligent and reasonably efficient manner. That's an issue of integration and data federation [and addressing things like] data integrity, access, privacy and security. It's really a very challenging area for any IT company."

Therein lies the biological science sector's appeal to IBM: these capabilities require the aggregation of many of the IT-related disciplines in which IBM has been working for years. These include areas such as data mining, for spotting patterns in large masses of data; powerful computing arrays made up of clusters of inexpensive Linux servers; security, for managing access to and use of the data; grid computing, for linking scientists with distant, powerful computers; data integration software such as IBM's DiscoveryLink, designed specifically for life sciences companies; and workflow, for speeding the flow of information and potentially shortening the time to approval for new drugs.

These technologies have generally been implemented separately according to general business requirements. But put together, the industry is like a proving ground - due to grow to be worth some $US40 billion by 2004, Jasinski says - where IBM's disparate efforts will come together into a focused solution. The result will be a powerful information management platform enabling the real-time collaboration that drives today's scientific communities.

IBM's eagerness to own this space, and its standing commitment to the life sciences sector, may well have given it a strong head start in legitimising the bioinformatics industry. Yet as the company's large-scale computing platforms continue to gain traction at the big end of town, it will also have to wrestle with the fact that the high cost of this technology will put these tools out of reach of the small, innovative companies that need it most.

Bridging this digital divide, will be a major priority for IBM as it seeks to commercialise grid computing technologies through involvement with ventures such as Australia's emerging GrangeNet project and similar networks overseas. By giving even small biotech companies access to these large-scale resources, Jasinski hopes IBM will be able to work in close partnership with an industry whose growth has stunned even him.

"What has astounded me about [the bioinformatics] industry is the speed at which it moves," he says. "Two years ago, everybody was talking about gene sequencing and genomics, and now it's a done deal. Now they're talking about gene expression and integrating medical data with genotyping data. All of this drives different ideas about what drives IT, and it's pushing us hard and fast to think about how to do this in an efficient way."