SAN FRANCISCO--Evolving trends in theory and technology were presented at the Cambridge Healthtech Institute’s Genome Tri-Conference here February 26-March 3. More than 100 speakers addressed 900 scientists and business developers at the event, which was composed of three sub-conferences: Genomic Opportunities, Commercial Implications, and Gene Functional Analysis. Approximately 60 service and equipment companies also presented in the exhibit hall. The first such meeting held here seven years ago had 300 attendees.
Recent innovations in theory and technology led several speakers to refer to "Moore’s Law of Genomics." Not only are costs dropping and speeds increasing in genomic analysis, as with silicon microprocessors, but the data are becoming more comprehensible, they said.
Observing the growth of genomics and bioinformatics as an industry, Charles Cantor, chief scientific officer of Sequenom, pointed to an initial public investment of $1.5 billion in the Human Genome Project, which has today generated a private sector worth perhaps 30 times that amount. "A challenge to all of us is to find the best way for the public and private sectors to complement each other," he added.
Referring to new ways of looking at output from the expression microarray, a five-year-old invention from Stanford University, David Botstein, chair of genetics there, said that as the field comes to terms with huge datasets, "we are going from hypothesis-limited back to hypothesis-driven science in genomics."
Lymphochips and brute force
Botstein described one new method called the "Lymphochip," an array of 18,000 genes that are used to profile heterogeneous breast cancer samples. Signatures produced by the device show that "cancer is more reproducible than we thought," he said. "We can reproduce the metastatic neighborhood, and in the future use this knowledge to lead us to interference-based treatment methods, as Judah Folkman has begun to do with angiogenesis." The Lymphochip and its "clustering by correlation" method is described in the February 3, 2000 issue of Nature.
Another advance both theoretical and technical in nature is the yeast two-hybrid system, which looks past nucleotide sequence to the protein-protein interactions, which actually define phenotype. Russell Finley, a professor of molecular medicine and genetics at Wayne State University, uses the system in an 8500 x 8500 matrix to map every interaction among the 72 million possible in pheromone-mediated mating in Drosophila. The ultimate goal is to map all protein interactions that occur in the organism.
Charles Cantor described how brute force approaches to genome data are giving way to challenging new problems in interpretation of individual genomic variability. "There are 12 million SNPs, which are the most common and functionally interesting human genetic variations. They represent a tremendous opportunity," he said. Since it’s not possible to study all of them, the short term goal is to identify the important ones. Sequenom’s mass spec/time-of-flight method, a rapid sequencing technology, is adaptable to this task, and will be particularly useful in measuring allele frequencies, he said.
New software, databases, and projects
Other presenters detailed related work. Mark Perlin, CEO of Cybergenetics in Pittsburgh, presented his mapping software, TrueAllele, an automated method for scoring microsatellite reads, which is said to eliminate a key bottleneck in the analysis of genetic data. Using an algorithm known as stutter deconvolution to eliminate artifacts called "PCR stutter bands" from the genomic data, manual intervention by the scientist is reduced and discovery is greatly speeded up, Perlin said.
Robert Strausberg of the US National Cancer Institute described his agency’s development of the Cancer Genetics Anatomy Project, a catalog of genes expressed during cancergenesis, which aims to interface genomics data with cancer research findings. An associated annotation initiative aims to spot SNPs in coding sequences. A summary of the project as well as spot chromosomal maps and full-length cDNAs are available at the institute’s web site, http://www.ncbi/nlm.gov/MGC. Ultimately, clinicians will be able to use molecular knowledge to discover "diseases within the disease," which are today lumped together because with light microscopy they have the same apparent morphology.
Yuri Nikolsky, from Integrated Genomics, a Chicago company, described his efforts to compile a database of metabolic and enzymatic pathways derived from large scale comparisons of genomic sequences of microbes important in disease and environmental interactions.
Andrew Conway of Silicon Genetics described his company’s "GeneSpring" line of software, now in version 3. Conway made the point that, unlike many startups, his company is profitable and has little debt, but is looking for collaborators to help it move into the high-throughput screening area.