Electronic Journal of Statistics
It is frequently of interest to identify simultaneous signals, defined as features that exhibit statistical significance across each of several independent experiments. For example, genes that are consistently differentially expressed across experiments in different animal species can reveal evolutionarily conserved biological mechanisms. However, in some problems the test statistics corresponding to these features can have complicated or unknown null distributions. This paper proposes a novel nonparametric false discovery rate control procedure that can identify simultaneous signals even without knowing these null distributions. The method is shown, theoretically and in simulations, to asymptotically control the false discovery rate. It was also used to identify genes that were both differentially expressed and proximal to differentially accessible chromatin in the brains of mice exposed to a conspecific intruder. The proposed method is available in the R package github.com/sdzhao/ssa.
Original Publication Citation
Zhao, S. D., & Nguyen, Y. T. (2020). Nonparametric false discovery rate control for identifying simultaneous signals. Electronic Journal of Statistics, 14(1), 110-142. https://doi.org/10.1214/19-EJS1663
Zhao, Sihai Dave and Nguyen, Yet Tian, "Nonparametric False Discovery Rate Control for Identifying Simultaneous Signals" (2020). Mathematics & Statistics Faculty Publications. 171.
Electronic Journal of Statistics is an open access journal. Copyright for all articles in EJS is CC BY 4.0.