Thursday, August 24

7:30 Registration and Morning Coffee

Gene Expression Analysis

8:00 Chair's Opening Remarks

Keynote Presentation

8:10 New Tools for a New Era in Genome Based Biomedical Research
Stephen E. Lincoln, Ph.D., Vice President, Informatics, Affymetrix 
Over the last decade, gene expression micorarrays have proven a critical component of many industrial and academic research programs. While this of course continues, microarray technology today allows us to interrogate cells with far greater resolution and precision than previously possible with any technology. Arrays now exist for interrogating complete RNA transcripts and thus elucidating splice variation. Similarly we can follow transcription of non-coding, and putatively regulatory RNA species. Combined with other arrays which assay genomic DNA, thus detecting copy number changes and SNP genotypes, these microarrays enable new paradigms in biomedical research and development.

8:50 Linking Sets of Genes to Biological Outcome
S. Stanley Young, Ph.D., Director, Bioinformatics, National Institute of Statistical Sciences
Microarray, proteomics, metabolomics, etc., all produce data sets where there are many more predictor variables than observations. There are correlations among these variables; indeed, 20,000 genes cannot all be active at the same time or marching to their own drummers. There is a need to utilize the correlations among these predictors to link outcomes, disease, drug effects, etc. to correlated sets of predictors. Inference methods are combined with matrix factorization to make this link. A public data set is used to show how sets of genes can be found and linked to biological outcome.

9:20 Technology Spotlights

9:50 Coffee Break, Poster and Exhibit Viewing

10:30 Solving the “RMA Database” Problem: Training a Reference RMA Model Based on Thousands of Biologically Diverse Affymetrix Samples
Mark Porter, MS, Director of Biostatistics and Data Analysis,Gene Logic Inc. 
Some of the most popular normalization and summarization methods for Affymetrix GeneChip arrays rely on multiple arrays to properly model probe effects (e.g., RMA, MBEI, PLIER). Traditionally, these models have been trained on and applied to samples from a single experiment and may not be appropriate for samples external to the experiment. One limitation of this workflow is that continuously updated archival expression databases are not possible unless a single reference model can be developed. We have developed such a reference RMA model based on thousands of biologically diverse samples from Gene Logic’s BioExpress reference database. The resulting model, called “refRMA”, can summarize independent samples of various organ and pathology types, while retaining the general characteristics of RMA. This talk will focus on the comparison of our refRMA model to that of classically trained RMA models as well as the types of applications that can benefit the microarray industry with such an approach.

11:00 Androgens Drive Divergent Responses to Salt Stress in Male Versus Female Rat Kidneys 
David L. Gerhold, Ph.D., Senior Research Fellow, Molecular Investigative Toxicology Department, Merck & Co.
Male rats and humans tend to retain excess dietary salt and increase blood pressure compared to females. Microarray studies in rats identified a surprising number of gender differences in kidney gene expression. Salt-sensitive male and female rats reacted to salt loading with virtually non-overlapping gene expression responses. We propose that male rats adapt to enable “fight or flight” responses, whereas female rats adapt to support gestation.

11:30 Panel Discussion Gene Expression Analysis

12:00 Lunch on Your Own or (Technology Workshop Sponsorships Available)

1:30 Chair's Remarks

1:35 Alternative Splicing Comes of Age: Tools for Detecting Variant Transcripts and Applications in Drug Discovery 
Richard Einstein, Ph.D., Vice President, Splice Array Products, ExonHit Therapeutics, Inc. 
A large number of RNA transcripts are produced from a much smaller set of genes and alternative splicing is a major mechanism responsible for generating this diversity. Alternative splicing mechanisms have been identified in many disease processes and therapeutics targeting this mechanism has started to appear in the literature, bringing this process to the forefront in drug discovery. In addition, analytical tools are now available for high-throughput detection of alternatively spliced transcripts and will facilitate future discoveries. This session will introduce the concepts and mechanisms as well as the biological importance of alternative splicing, and the application in biomedical research. 

2:05 Profiling Alternatively Spliced mRNA Isoforms for Prostate Cancer Classification
Chaolin Zhang, BE, Cold Spring Harbor Laboratory /Department of Biomedical Engineering, SUNY at Stony Brook
Gene expression microarrays have been widely used to identify cancer markers. These studies assume the old dogma of one gene, one mRNA, which may underestimate complexity of tumorigenesis. This talk presents a study of using splicing-isoforms for prostate cancer classification. Integrating the information of alternative splicing allows the identification of new marker candidates and improvement in classification performance. These results suggest the unique invaluable information from splicing-isoform profiling, which cannot be provided by the traditional gene expression microarrays.

2:35 Knowing What You Don't Know: Probe Level Analysis of Exon Array Data
Hugh Salamon, Ph.D., Senior Computational Biologist, Department of Computational Biology, Berlex Biosciences
Individual probe hybridization intensities measured on the HuEx 1.0 ST Affymetrix platform vary greatly in their intensity and their ability to discriminate mRNA levels, even for probes which are sequence mapping neighbors targeting the same transcripts. In the first example study examined, we found that the summarization method PLIER provided data which, after T-tests and false discovery rate corrections, had no power to detect differential expression at the exon, transcript, or entire locus levels. A probe-level statistical analysis instead provided evidence for hundreds of differentially expressed loci at a false discovery rate of 0.05. Appropriate statistical modeling will be needed to ensure that less responsive probes do not adversely impact detection of alternative splicing.

3:05 Refreshment Break, Poster and Exhibit Viewing

3:45 Characterization of the Stem Cell Transcriptome using Computational Analysis and Large-Scale Genetic Screens
Moshe Pritsker, Ph.D., Research Fellow, Neurosurgery, Harvard Medical School; Massachusetts General Hospital
Rational design of future stem cell therapies requires comprehensive identification and functional annotation of the stem cell transcriptome. Toward this goal, we have conducted computational and experimental analyses to characterize alternative splicing variants generated in hematopoietic and embryonic stem cells. We have also developed a high-throughput microarray-assisted approach for systematic gain-of-function identification of genes regulating stem cell self-renewal and differentiation. Biological insights obtained and perspectives for possible applications in development of new therapies will be discussed.

4:15 Alt-Splice ANOVA - A Statistical Test to Detect Alternative Splicing
Thomas J. Downey, MS, President, Partek, Inc. 
This talk demonstrates how alternative splicing events can be detected using a mixed model analysis of variance (ANOVA). The benefits of this approach are numerous. The method not only produces a formal p-value for each gene which represents the probability that the gene is exhibiting alternative splicing, but also produces a p-value for differential expression at the gene-level. In addition the method is very flexible and allows a variety of different alternative splicing patterns to be detected.

4:45 Panel Discussion Alternative Splicing - Are We Making Biological Sense?

5:15 Pizza and “Micro” Brews Networking Reception

6:30 Close of Day