Hidden Markov models for simultaneous testing of multiple gene sets and adaptive and dynamic adaptive procedures for false discovery rate control and estimation

Abstract

Storey (2002) advocated the approach of conservative point estimation of false discovery rate (FDR) and showed that it is closely related to the classical Benjamini and Hochberg (1995) procedure that controls FDR. Storey et al. (2004) further proved that a related procedure strongly controls FDR through the elegant use of empirical processes. Since then, many FDR control procedures have been developed. In light of these new developments, we continue Storey’s work on the conservative point estimation of FDR. We first give a corrected version of the finite sample proof for conservative point estimation of FDR. Then we survey existing theoretically proven FDR control procedures and show that some of these procedures can be cast as plug-in procedures with their estimators of the proportion of true null hypotheses (π0) more conservative