Power and Stability Comparisons of FDR Controlling Multiple Testing Procedures

• Activity Number: 480
• Type: Contributed
• Date/Time: Wednesday, August 3, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Dongmei Li*+
• Companies: University of Hawaii
• Keywords: Multiple testing ; FDR ; Power ; Stability

Abstract

Microarray analysis has been widely used to identify differentially expressed genes under different treatment conditions or to identify gene markers associated with certain disease or quantitative traits. False discovery rate (FDR) is considered to be a proper type I error rate to control for microarray data analysis. Various multiple testing procedures have been proposed to control the FDR. However, the stabilities of those multiple testing procedures have not been evaluated yet. To determine the procedure that most likely produces reproducible results, simulation studies are conducted to examine the stabilities of five frequently used multiple testing procedures (Benjamini and Hochberg, 1995; Benjamini and Yekutieli, 2001; Storey, 2002; Storey et al., 2004; Benjamini et al., 2001) with FDR control. Based on the performance of the five multiple testing procedures on FDR controlling, power, and stability, the Benjamini and Hochberg (1995) and Benjamini et al. (2001) procedures seem to be good choices for microarray analysis. The leukemia and ovarian cancer microarray studies are used to illustrate the power and stabilities of these five FDR controlling multiple testing procedures.