Overcoming Adverse Effects of Correlations in Microarray Data Analysis

• Activity Number: 70
• Type: Contributed
• Date/Time: Sunday, August 1, 2010 : 4:00 PM to 5:50 PM
• Sponsor: Section on Statistical Computing
• Author(s): Linlin Chen*+ and Alexander Gordon and Galina Glazko
• Companies: Rochester Institute of Technology and University of North Carolina Charlotte and University of Rochester
• Keywords: microarray ; multiple testing procedure ; correlation ; type I and II error

Abstract

Due to the existence of extremely strong and long-ranged correlations between expression levels of different genes, all procedures which are commonly used to detect the genes differentially expressed between two or more phenotypes are unable to overcome the two main problems: high instability of the number of false discoveries and low power. It may be impossible to completely understand these correlations due to the complexity of their biological nature. We have proposed a new multiple testing method to balance type $I$ and type $II$ errors in an optimal, in a sense, way. However, the correlation structure of microarray data is still the main obstacle standing in the way of this and other gene selection procedures. To remove this obstacle, we further improve the statistical methodology by exploiting the remarkable property of low dependency between the terms of the so-called $\delta$-se