A0160
Title: Structure adaptive multiple testing
Authors: Xianyang Zhang - Texas A\&M University (United States) [presenting]
Jun Chen - Mayo Clinic (United States)
Abstract: Conventional multiple testing procedures often assume that the hypotheses for different units are exchangeable. However, in many scientific applications, external structural information regarding the patterns of signals and nulls are available. We introduce new multiple testing procedures that can incorporate various types of structural information including (partial) ordered structure, smooth structure, group/multi-group structure and covariate-dependent structure. We develop an EM-type algorithm to efficiently implement the proposed procedures and justify the asymptotic validity of our method as the number of hypotheses goes to infinity. We investigate the finite sample performance of the proposed method through extensive simulation studies and real data analysis and find that the new approach is highly competitive to the state-of-the-art approaches in the literature.
