gbh: gbh: Grouped Benjamini Hochberg

In nignatiadis/IHWpaper: Reproduce figures in IHW paper

Description

gbh: Grouped Benjamini Hochberg

tst_gbh: wrapper for gbh with method="TST" lsl_gbh: wrapper for gbh with method="LSL"

Usage

gbh(unadj_p, groups, alpha, method = "TST", pi0_global = "weighted_average")

tst_gbh(unadj_p, groups, alpha, ...)

lsl_gbh(unadj_p, groups, alpha, ...)

Arguments

unadj_p	Numeric vector of unadjusted p-values.
groups	Factor to which different hypotheses belong
alpha	Significance level at which to apply method
method	What pi0 estimator should be used (available "TST","LSL")
pi0_global	GBH requires also a pi0 estimate for the marginal p-value distribution. Can either apply pi0 estimation method to all p-values (pi0_global="global") or use a weigted averarage (pi0_global="weighted_average") of the pi0 estimates within each stratum. This is not explicitly stated in the paper, but based on a reproduction of their paper figures it seems to be the weighted_average.
...	Additional arguments passed from tst_gbh/lsl_gbh to gbh

Value

GBH multiple testing object

Functions

• tst_gbh: Wrapper of GBH with TST pi0 estimator

• lsl_gbh: Wrapper of GBH with LSL pi0 estimator

References

Hu, James X., Hongyu Zhao, and Harrison H. Zhou. "False discovery rate control with groups." Journal of the American Statistical Association 105.491 (2010).

Examples

     sim_df <- du_ttest_sim(20000,0.95, 1.5)
     sim_df$group <- groups_by_filter(sim_df$filterstat, 20)
     obj <- tst_gbh(sim_df$pvalue, sim_df$group, .1)
     sum(rejected_hypotheses(obj))

nignatiadis/IHWpaper documentation built on Jan. 18, 2021, 2:24 p.m.