Overview
Services
Experimental Design
Statistical Analysis
References
Contact

Overview
The Functional Genomics Core Laboratory offers "full-service" bioinformatics analyses of microarray-generated data. This bioinformatics support unit develops and maintains a gene expression database and manages a bioinformatics and statistical software suite. Members also provide on-site user support, conduct tutorials and workshops on microarray data analysis, and provide comprehensive data management and analysis.

Services
Specific services include, but are not limited to:

• Consultation on experimental design
• General statistical consulting
• Data management and preparation (e.g., pre-processing)
• Custom software design and development
• Biostatistics and bioinformatics support for all types of microarray platforms: two-color cDNA, one-color and two-color oligonucleotide, Affymetrix
• Microarray data normalization and identification of differentially expressed genes
• Microarray data visualization and exploratory analysis (e.g., clustering)
• Data analysis, interpretation, and report preparation
• Assistance on grant or protocol submission (e.g., preparing statistical analysis plan or database description)
• Manuscript review and preparation
• Advanced training on microarray analysis

Experimental Design
There are many choices possible for microarray experimental designs, with each choice having advantages and disadvantages. For example, for two-color microarrays, designs belong to a type of experimental designs known in the field of statistics as incomplete block designs[1]. The experimental layout of a microarray study has a substantial impact on the precision with which the effects of interest can be estimated. Therefore, the Core provides experimental design analysis to identify a specific design that is suitable for addressing a study’s scientific goals. For example, to obtain unbiased estimates of relative gene expression, a microarray design must ensure that ancillary experimental effects can be accounted for. In addition, a good microarray design will make the most of expensive resources by maximizing precision while controlling costs. Design robustness is another important concern. Other design decisions include how to account for biological variability, and whether to pool mRNAs.

Statistical Analysis
Statistical analysis and data normalization are carried out with several software packages:
Spot-On software developed at the University of Washington by Dr. Roger Bumgarner[2]
Spot, a software package for the analysis of microarray images from CSIRO, Australia's Commonwealth Scientific and Industrial Research Organisation
MicroArray Analysis of Variance (MA-ANOVA) software developed by the Statistical Genetics Group at the Jackson Laboratory[3]
Bioconductor software developed by collaborators based at the Biostatistics Unit of the Dana Farber Cancer Institute at Harvard Medical School/Harvard School of Public Health[4]
GeneSpring™, produced by Silicon Genetics, Redwood City, California
GeneTraffic ®, produced by Iobion Informatics, La Jolla, CA .
The Spot-On software is freely available to UW researchers. The Spot software has been purchased and is available. The MA-ANOVA and the Bioconductor packages are freely available to the academic community.
Spot-On provides tools for image analysis of scanned microarrays and a method of non-linear normalization. The software can also generate lists of differentially expressed genes in simple dye-flip experiments by an error analysis based on replicated spots. Spot is based on a seeded region growing algorithm that allows for very fast image analysis with minimal user input. MA-ANOVA provides additional options for normalization and data transformations. The outstanding feature of MA-ANOVA is the ability to perform the analysis of variance on microarray data with methods of non-parametric inference. This allows investigators to use more sophisticated experimental designs in their microarray studies. MA-ANOVA also offers a method to incorporate error-modeling into higher-order analyses such as cluster analysis. The Bioconductor software provides an additional set of tools for data normalization and visualization. GeneTraffic provides tools for data management, data analysis, cluster analysis, cluster visualization, and mechanisms to annotate gene lists so that investigators can begin to explore gene families and biological pathways. GeneSpring® provides tools for data normalization, cluster analysis and mechanisms to annotate gene lists so that investigators can begin to explore differential gene expression across groups of genes as well as biological pathways.

References

1. Kerr and Churchill. 2001. Experimental design for gene expression microarrays. Biostatistics 2:183-201.
2. Geiss GK, Bumgarner RE, An MC, Agy MB, van’t Wout AB, Hammersmark E, Carter VS, Upchurch D, Mullins JI, Katze MG. 2000. Large-scale monitoring of host cell gene expression during HIV-1 infection using cDNA microarrays. Virology 266:8-16.
3. http://www.jax.org/research/churchill
4. http://www.bioconductor.org/
5. Kerr and Churchill. 2001. Bootstrapping Cluster Analysis: Assessing the Reliability of Conclusions from Microarray Experiments. PNAS 98:8961-8965.

Contact

Theo Bammler, PhD, Bioinformatics Coordinator
tbammler@u.washington.edu, (206) 206-543-2061
UW Box 354695

Dick Beyer, PhD, Research Scientist
dbeyer@u.washington.edu, (206) 616-7378
UW Box 354695