BOSTON — Boston and its brainy neighbor Cambridge, Mass., served as appropriate hosts last week to conferences at opposite ends of the bioinformatics spectrum.
While academic researchers discussed the newest crop of computational biology algorithms at RECOMB 2005, held on the Massachusetts Institute of Technology campus in Cambridge, their commercial counterparts were schmoozing it up trade-show style at Bio-IT World across the Charles River at Boston's Hynes Convention Center.
A few attendees made it to both meetings, but participants at each conference were for the most part utterly unaware of events taking place across the river — perhaps a fitting metaphor for the yawning gap that still exists between academic bioinformatics development and the commercial informatics sector.
Granted, the meetings have very different scopes: While RECOMB focuses strictly on emerging methods in computational biology, Bio-IT World casts a much broader net, providing a showcase for hardware and storage vendors, clinical trials software, knowledge-management solutions, and other IT offerings of interest to the life science and healthcare industry.
But some of those vendors may have been interested to learn a bit about the new methods being showcased just a few miles away. Who knows? Maybe a few of these algorithms will be part of their commercial product offerings a year or two from now.
RECOMB: Way Beyond Sequence Alignment
The ninth annual Research in Computational Molecular Biology conference kicked off on May 15 with a series of talks on network analysis, which is proving to be an area of growing interest for computational biologists. New methods are rapidly appearing on the horizon in this field, as illustrated by the work of Perdue University's Mehmet Koyut rk, who described a new pairwise alignment method for pathway interaction networks.
Like PathBlast, one of the first algorithms of this type [BioInform 02-28-05], which was also discussed at RECOMB this year, Koyut rk's method uses protein homology between different species as a starting point for aligning the networks, but builds on previous methods by using information on evolutionary patterns — such as gene duplications and the insertion or deletion of interactions — to score the alignment.
Another area of rapid development among the RECOMB crowd is in RNA analysis. Work at this year's meeting fell into two camps: prediction of secondary structure, and prediction of targets for RNA interference. Shaojie Zhang of the University of California, San Diego, discussed RNAscf (RNA Stack-based Consensus Folding), a method for performing multiple alignment and structure prediction for RNA molecules at the same time. Chaya Zilberstein of Technion, meanwhile, presented a tool called miRNA Xpress that predicts the mRNA targets for microRNAs in plants.
Computational biologists are also working to improve mass spectrometry analysis. Several methods presented in the first few days at RECOMB claim to offer better or faster results than Mascot and Sequest, the current workhorses in the field. Ari Frank of UCSD discussed a new approach called Inspect that uses a "filtration" approach to speed the searching process — much like the heuristics that enable Blast to run faster than Smith-Waterman. Frank said that the algorithm offers a 150-fold speedup over Sequest. A paper describing the method is in press.
Bio-IT World: Semantic Web, PGx, and Oceans of Data
Meanwhile, in Boston, attendees of the Bio-IT World conference got a taste of several high-level initiatives that may have some impact in the commercial informatics sector, either in the long or short term.
Keynotes from Tim Berners-Lee, Lawrence Lesko, and Craig Venter provided an overview of the semantic web, the FDA's pharmacogenomics activities, and the emerging science of environmental sequencing, respectively.
Of the three, Berners-Lee's talk generated the most buzz on the conference floor, as attendees debated whether the man credited with inventing the Internet will have the same success with the semantic web, which some in the life science informatics community view as a promising way to solve the field's persistent data-integration problems [BioInform 11-08-04].
Describing the semantic web as "totally a grassroots thing," Berners-Lee encouraged people to start adopting available technologies, such as RDF and OWL, immediately. "Don't ask your boss," he said, "tell him afterwards."
The life science sector "is crying out to be the flagship area for the semantic web to take off," he said, citing some prototype efforts in the life science community, such as BioDash [http://www.w3.org/2005/04/swls/BioDash/Demo/] as an early success story.
While Berners-Lee's talk may have been a bit too blue-sky for some attendees, Lesko's overview of the FDA's work in pharmacogenomics offered some immediate challenges for the informatics sector.
Lesko, director of the Office of Clinical Pharmacology and Biopharmaceuticals at the FDA's Center for Drug Evaluation and Research, cited data interpretation as one of the primary barriers to widespread adoption of pharmacogenomics in clinical practice.
Lesko said the recent approval of the AmpliChip CYP450 test has led to a quandary for some of its early physician adopters, who are running the test and then asking what the data means, and whether they should adjust their dosage. While the AmpliChip has the potential to reduce adverse drug events by 20 percent, he said its ultimate utility is still an "open question" until researchers and clinicians gain more experience — and some help analyzing the data.
On the final day of the conference, Craig Venter discussed the massive amounts of data he is gathering in his various environmental genomics projects — both by sea and by air. The J. Craig Venter Institute has developed a number of new bioinformatics methods to analyze the millions of microbial sequence reads the effort is generating, but Venter was unwilling to disclose further details of those methods before they are published later this year.
(See "News Briefs from the Bio-IT World Conference Floor," this issue, for news items from the show.)
— Bernadette Toner ([email protected])