Empirical Evaluation of Methodologies for Microarray Data Analysis, With Some Thoughts on Statistical Implications

Tuesday, September 30, 2003 - 11:30am - 11:55am
Keller 3-180
Kathleen Kerr (University of Washington)
This talk will present recent results of empirical tests of various methodologies of microarray data analysis. The data are spike-in assays produced as part of a standardization experiment performed by the Toxicogenomics Research Consortium. Findings support the use of an intensity-based normalization procedure and provide strong evidence that the practice of local background subtraction is detrimental. Statistically, the most interesting findings pertain to the relative ability of various methodologies for detecting differentially expressed genes. The speaker will present these results along with some opinions on directions of research to advance statistical methodology for microarray data analysis.