You are here:
MICROARRAY DATA ANALYSIS USING MULTIPLE STATISTICAL MODELS
Citation:
Bao, W, J E. Schmid, A K. Goetz, M. Ouyang, W. J. Welsh, AND A. I. Brooks. MICROARRAY DATA ANALYSIS USING MULTIPLE STATISTICAL MODELS. Presented at Bioinformatics and Genome Research Conference, San Francisco, CA, June 21 - 24, 2004.
Description:
Microarray Data Analysis Using Multiple Statistical Models
Wenjun Bao1, Judith E. Schmid1, Amber K. Goetz1, Ming Ouyang2, William J. Welsh2,Andrew I. Brooks3,4, ChiYi Chu3,Mitsunori Ogihara3,4, Yinhe Cheng5, David J. Dix1. 1National Health and Environmental Effects Research, Laboratory, Office of Research and Development, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711.2 Department of Pharmacology, Robert Wood Johnson Medical School and Informatics Institute, University of Medicine and Dentistry of New Jersey, NJ 08854.3Center for Functional Genomics,4Department of Environmental Medicine,5Department of Computer Science, University of Rochester School of Medicine and Dentistry, Rochester, NY, 14642.
Microarray technology has become an invaluable tool for functional genomic studies in many biological fields. The enormous amount of data from DNA microarrays, the absence of a generally accepted analysis method, and the uncertainty in what constitutes a ?correct? result, make data analysis very challenging and comparison of different analyses even more difficult. Various data analysis modelsgenerate diverse results due to differencesin data normalization, transformation and analytical approaches. To compare and to gain more confidence in the significant differentially expressed genes from a DNA microarray study, we analyzed a dataset using four independent software packages and statistical models. The dataset was derived from a reproductive toxicology study involving exposure of mice to conazole fungicides. The four analysis approaches included Practical and Sequential Strategies (PASS)for microarray data processing and analysis (Bao et al., 2004.EHP Toxicogenomics,submitted), Genespring 6.1 (http://www.genespring.com/), Independent Consistent Evaluation Discriminator (ICED)(http://fgc.urmc.rochester.edu/data_analysis.html), and a Bayesian model for analysis of microarray expression data (Baldi, P., and A. Long, 2001.Bioinformatics17:509-19). Lists of differentially expressed genes were generated by each of these four methods, and the different and common genes from the four models were compared and analyzed. Additionally, the biological function and toxicity pathways common to genes from the four models were characterized usingIngenuity Pathways Analysis package (http://www.ingenuity.com/). These comparisons revealed a core set of differentially expressed genes, biological functions, and toxicity pathways identified by multiple analysis methods, independent of the methods used to analyze the microarray data.
This is an abstract of a proposed presentation and does not necessarily reflect EPA policy.