rth-Rank Ordered Statistic for Microarray Meta-Analysis

• Activity Number: 70
• Type: Contributed
• Date/Time: Sunday, August 1, 2010 : 4:00 PM to 5:50 PM
• Sponsor: Biometrics Section
• Author(s): Chi Song* and George Tseng+
• Companies: University of Pittsburgh and University of Pittsburgh
• Keywords: meta-analysis ; microarray ; expression profile ; rth-rank ; vote-counting

Abstract

As the use of microarray in biomedical area is more prevalent, numerous datasets have been generated from different studies. Meta-analysis methods are of great demand to compare and synthesize studies of similar biological interests. In the biological literature so far, several variants of vote-counting method has been widely used though such a method is well-known to be inferior, in terms of less statistical power. In this paper, we explore the use of rth-rank p-value (rth-Rank) in meta-analysis. We show that rth-Rank is a generalization of minP/maxP or the traditional vote-counting method. The advantages of this extended statistic can be shown by simulations and by two genomic meta-analysis applications on diabetes and depression, in which the studies are heterogeneous and have small effect sizes. The results show that rth-rank method provides better statistical power and robustness.