Biostatistics Weekly Seminar

Introduction to False Discovery Rates in Large-Scale Inference

Valerie Welty
Vanderbilt University School of Medicine

The false discovery rate as defined by Benjamini and Hochberg has become a popular alternative to the family-wise error rate (FWER) in error control when testing a large number of hypotheses (i.e. large-scale inference). The type of false discovery rate (FDR) used by Benjamini and Hochberg is one of several potential FDR definitions studied in the last few decades, with considerable contributions by Storey et al. and Efron et al. Use of these alternative FDRs, such as the positive false discovery rate (pFDR), move the focus from testing algorithms to estimation of FDRs. A variety of non-parametric and parametric Empirical Bayes approaches for estimating probabilistic FDRs based on a large number of observed test statistics have been developed. In this seminar, I will introduce the relevant components of large-scale inference, review the various FDRs, outline the connections between testing procedures and false discovery rate estimation, and briefly introduce several Empirical Bayes approaches for FDR estimation focusing on the methods of Efron et al. .

MRBIII, Room 1220
20 November 2019

Speaker Itinerary

Topic revision: r1 - 18 Nov 2019, TawannaPeters

This site is powered by FoswikiCopyright © 2013-2020 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback