ihw_naive: IHW wrappers

In nignatiadis/IHWpaper: Reproduce figures in IHW paper

Description

IHW wrappers

Usage

ihw_naive(unadj_p, filterstat, alpha)

ihw_ecdf_5fold(unadj_p, filterstat, alpha)

ihw_5fold(unadj_p, filterstat, alpha)

ihw_5fold_reg(unadj_p, filterstat, alpha)

ihw_bonf_5fold_reg(unadj_p, filterstat, alpha)

ihw_storey_5fold(unadj_p, filterstat, alpha)

Arguments

unadj_p	Numeric vector of unadjusted p-values.
filterstat	Factor to which different hypotheses belong
alpha	Significance level at which to apply method

Details

These are closures, which apply IHW with custom prespecified parameters. These correspond to interesting settings, for which it is convenient to be able to immediately call the corresponding functions, rather than having to specify parameters each time. Thus they make it easier to benchmark. All of these wrappers are defined in 2 lines of code, so the settings pertaining to each one can be inspected by typing the functions name into the console.

Value

ihwResult multiple testing object

Functions

• ihw_naive: IHW naive

• ihw_ecdf_5fold: IHW (E2) with 5 folds

• ihw_5fold: IHW (E1-E2) with 5 folds

• ihw_5fold_reg: IHW (E1-E2-E3) with 5 folds

• ihw_bonf_5fold_reg: IHW-Bonferroni (E1-E2-E3) with 5 folds

• ihw_storey_5fold: IHW (E1-E2) with 5 folds and Storey's pi0 estimator

Examples

     sim_df <- du_ttest_sim(20000,0.95, 1.5)
     obj <- ihw_5fold(sim_df$pvalue, sim_df$filterstat, .1)
     sum(rejected_hypotheses(obj))

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