An Adaptive Two-Step Procedure to Control Mean Number of False Discoveries

• Activity Number: 403
• Type: Topic Contributed
• Date/Time: Tuesday, August 3, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Dongmei Li*+
• Companies: The University of Hawaii at Manoa
• Keywords: Bonferroni ; Adaptive two-step procedure ; Mean number of false discoveries

Abstract

When Bonferroni procedure is used to control the mean number of false discoveries in microarray data analysis, it has an equivalent power and a superior stability in terms of the variance of both true discoveries and total discoveries compared to Benjamini-Hochberg procedure (Gordon et al., 2007). In this work, an adaptive two-step procedure developed from the Bonferroni procedure is present, which can control the mean number of false discoveries at a desired level. Compared to the Bonferroni procedure, simulation studies show that this adaptive two-step procedure has increased powers and decreased variances of both true discoveries and total discoveries. The above advantages of the adaptive two-step procedure increase as the proportions of true null hypotheses decrease under both independent and dependent cases.