A weighted FDR procedure under discrete and heterogeneous null distributions
Abstract
Multiple testing with false discovery rate (FDR) control has been widely conducted in the ``discrete paradigm" where p-values have discrete and heterogeneous null distributions. However, in this scenario existing FDR procedures often lose some power and may yield unreliable inference, and for this scenario there does not seem to be an FDR procedure that partitions hypotheses into groups, employs data-adaptive weights and is non-asymptotically conservative. We propose a weighted FDR procedure for multiple testing in the discrete paradigm that efficiently adapts to both the heterogeneity and discreteness of p-value distributions. We theoretically justify the non-asymptotic conservativeness of the weighted FDR procedure under independence, and show via simulation studies that, for multiple testing based on p-values of Binomial test or Fisher's exact test, it is more powerful than six other procedures. The weighted FDR procedure is applied to a drug safety study and a differential methylation study based on discrete data, where it makes more discoveries than two existing methods.
Cite
@article{arxiv.1502.00973,
title = {A weighted FDR procedure under discrete and heterogeneous null distributions},
author = {Xiongzhi Chen and R. W. Doerge and Sanat K. Sarkar},
journal= {arXiv preprint arXiv:1502.00973},
year = {2019}
}
Comments
29 pages; in this version: (1) new weights from arXiv:1812.06551v2 are employed by the wFDR procedure to ensure its conservativeness under independence; (2) non-asymptotic upper bound is provided for the wFDR procedure that uses plug-in weights that are induced by the proportion estimator of arxiv:1410.4274v2; (3) simulation study under dependence uses more a complicated block dependence structure