Each year, the invited speakers at the International Conference on Research in Computational Molecular Biology challenge the gathering of computer scientists, statisticians, and mathematicians with some of biology’s most complex problems, but this year’s meeting may have been the first time attendees were literally reprimanded by a speaker.
The meeting kicked off with a wake-up call from Craig Venter, who scolded the crowd for failing to oppose the “shady statistical arguments” used in a recent attack on Celera Genomics’ whole-genome shotgun technique.
“It’s an embarrassment that there has not been more of a response from all of you,” said the former president of Celera during the meeting’s annual Stanislaw Ulam Memorial Lecture. Referring to a recent paper published in the Proceedings of the National Academy of Sciences in which scientists from the human genome project criticized Celera’s use of public data, Venter told the more than 500 computational scientists in the audience that they “should be offended by the [paper’s] use of mathematics to fool people.”
Venter’s impassioned plea to the logician-heavy crowd highlighted the contrast in cultures between biology and computer science that has become RECOMB’s hallmark: Over the course of the four-day event, 35 papers on new algorithms and statistical methods were broken up by eight invited speakers who delivered a higher-level vision of current research in biology.
“This is a real geek meeting,” noted program chair Gene Myers proudly before introducing Venter. “We haven’t even gotten to the first talk and there have been two diagrams.” Myers wasn’t exaggerating: He was responsible for the first, a color-coded bar graph detailing the various events scheduled during the conference, which took place April 18-21 in Washington, DC. The second was a detailed map of the conference venue.
But while enough diagrams, formulas, tables, and graphs followed to satisfy even the most hard-core geeks in attendance, most participants agreed that the invited speakers were the highlight of the conference.
Marc Vidal of the Dana Farber Cancer Institute discussed his lab’s work on generating a complete protein-protein interaction map for Caenorhabditis elegans. “Even imperfect maps can be useful,” said Vidal, noting that the full value of such mapping efforts won’t be evident until they are integrated into a “proteome atlas” along with other functional maps, such as expression and phenotypic profiles. In building the interactome map, Vidal and his colleagues cloned 19,000 open reading frames to map protein function. In the process, they discovered that 28 percent of computationally predicted genes in the C. elegans genome needed “correction,” a point that Vidal said doesn’t bode well for annotation of the human genome.
Harold “Skip” Garner proved to be an appropriate choice for the meeting’s Distinguished New Technologies Lecture as he shared the technology generated by his fertile research group at the Center for Biomedical Inventions at the University of Texas Southwestern Medical Center. In addition to a playfully named suite of bioinformatics software tools (including POMPOUS, SNIDE, ARROGANT, TRITE, and ARGH), Garner’s lab has developed a new approach to synthesizing microarrays using Texas Instruments’ digital light processor chip. Most of the lab’s tools are available at http://innovation.swmed.edu.
Placing computational approaches within the real-world context of rational drug design, Ruben Abagyan of the Scripps Research Institute discussed the advantages of computational structural proteomics. “Modeling can bridge the sequence-structure gap. Crystallization can’t,” said Abagyan. Scripps has been perfecting a modeling program based on internal coordinate mechanics to predict protein structures and perform virtual ligand screening. Abagyan said the method could predict the correct docking geometry 93 percent of the time.
Rockefeller University’s David Ho delivered the meeting’s Distinguished Biology Lecture, a review of his lab’s recent research on HIV replication in lymphocyte populations. In what he described as his “first talk for mathematicians and computer scientists,” Ho explained how his lab was able to resolve a controversy within the HIV research community over the decline in T-cells in HIV infection. The group tracked lymphocyte turnover in normal and HIV-infected patients in order to conclude that the drop-off was due to a high cell death rate rather than a low reproduction rate.
Another biology-focused talk came from Ali Brivanlou, also of Rockefeller, who discussed his group’s use of microarrays to study neural cell development in embryos of the frog Xenopus laevis. Specifically, his lab is studying the cell-cell interactions that occur during embryonic development and determine how cells “know” their identity. “For biologists, microarrays are more than statistical tools,” Brivanlou reminded his audience, pointing out that his lab used microarrays to find genes of previously unknown function that are only expressed during the first few hours of life in X. laevis. Brivanlou argued that many more genes of unknown function may play a key role in early development.
Moving back to computational biology, Evan Eichler of Case Western Reserve School of Medicine shared the approaches his lab developed to identify recent duplicates within the human genome. Eichler said his approach suggests that a number of important human genes will not possess orthologs in the genomes of other model organisms, a key point to keep in mind for accurate annotation of the final human genome draft sequence.
A final look at the challenges of genome annotation came from Gerry Rubin of the University of California, Berkeley, who pointed out that despite advances in computational science, experimentation will remain a necessary component of the “finishing” process for any genome. He detailed the efforts of the Berkeley Drosophila Genome project to combine experimental and computational biology in its annotation work.
Further RECOMB highlights included two award presentations: Adam Siepel of the University of California, Santa Cruz, went home with a brand-new IBM laptop as the winner of the best student paper and Sylvia Spengler won the Fred Howes Distinguished Service Award for her work as director of the National Science Foundation’s Biological Databases and Informatics program.
RECOMB 2003 will be held in Berlin, Germany, April 10-13, 2003.