This year’s TIGR Conference on Computational Genomics covered a broad spectrum of topics, but the common thread running through each talk was “a close tie to biology,” according to conference co-chair Tony Kerlavage.
Kerlavage, senior director of product development at Celera Genomics, said the organizers of the annual meeting are selective about the topics covered, seeking techniques and methods with direct medical relevance and applicability to “real-world problems.”
A key example at this year’s meeting, held in Baltimore, Md., November 28- December 1, was Roger Reeves’ discussion of his use of comparative genomics in Down syndrome research. Reeves, a professor in the Johns Hopkins School of Medicine department of physiology, described how comparisons between the human and mouse genome sequence could improve predictions of essential genes.
Reeves warned listeners that the NIH could use a bit of prodding to continue funding the mouse genome through completion. “We really need finished sequence to do any biological work,” said Reeves — a statement that brought cheers from conference attendees.
Richard Mural, director of scientific content and analysis at Celera, followed Reeves with another look at the new insights that nearly finished genomes are bringing to biological research. “It’s hard to remember this data has only been available for less than a year,” said Mural, flashing a PowerPoint slide of aligned — and nearly identical — mouse and human sequence. Mural suggested that because the two genomes are so similar, additional sequence from other vertebrates might be used as a “tiebreaker” between the two in cases where the correlation between conserved regions and vital functions is questionable.
One hot topic was the series of talks on phylogeny and evolutionary biology that kicked off the meeting. A keynote address from the University of Arizona’s Howard Ochman on “the expanding and contracting” E. coli genome initiated a lively debate on the still-controversial topic of horizontal gene transfer.
Another area of contention is in gene expression analysis. Several conference attendees remarked that the field is currently stalled on comparisons between clustering algorithms that are often indistinguishable from one another. The growing consensus is that the clustering algorithm used doesn’t really matter — it’s what follows that stage of analysis that brings biological knowledge in the form of predictive modeling, pathway prediction, or other context-based approaches.
On the data resource side, Owen White from TIGR and Ewan Birney from the European Bioinformatics Institute brought listeners up to date on the latest features available from the Comprehensive Microbial Resource and Ensembl, respectively. White said an upcoming service from CMR would enable users to submit sequence that TIGR will run through its annotation engine. Birney said the Ensembl team expects to have the mouse, ciena, tetraodon, and two fugu genomes available in early 2002, with the rat, zebrafish, and pig coming online later in the year.
The meeting’s two tutorials, one on gene finding methods by Georgia Tech’s Mark Borodovsky and another on IT infrastructure for high-throughput informatics by Innova IT’s Bruce Moxon, gave attendees an in-depth look at both ends of the computational genomics spectrum — from algorithm development to hardware trends.
Like many industry gatherings since September 11, attendance was lower than expected at this year’s conference. Kerlavage, who co-chaired the meeting with TIGR’s Steven Salzberg, said there were around 250 participants this year compared to 380 a year ago, and noted a marked decrease in international attendance compared to previous years.
But next year’s meeting will pick up again, Kerlavage said. And what will be the hot topics there? Regulatory regions, pathways, and networks, he predicted.