fdrtool_subset: Estimate (Local) False Discovery Rates For Diverse Test...
In ludgergoeminne/MSqRob: Robust statistical inference for quantitative LC-MS proteomics

fdrtool_subset

R Documentation

Estimate (Local) False Discovery Rates For Diverse Test Statistics

Description

This function is largely based on the fdrtool function from the fdrtool package by Korbinian Strimmer (http://strimmerlab.org). fdrtool takes a vector of z-scores (or of correlations, p-values, or t-statistics), and estimates for each case both the tail area-based Fdr as well as the density-based fdr (=q-value resp. local false discovery rate). The parameters of the null distribution are estimated adaptively from the data (except for the case of p-values where this is not necessary). Our only modification is adding the possibility to provide a reduced input (e.g. an input of which the excess in p values equal to 1 are removed) for estimating the null distribution while still correcting for the total number of input values.

Usage

fdrtool_subset(x, x_red = x, statistic = c("normal", "correlation",
  "pvalue"), verbose = TRUE, cutoff.method = c("fndr", "pct0", "locfdr"),
  pct0 = 0.75)

Arguments

`x`	A vector of the observed test statistics.
`x_red`	A subset vector of x that will be given to the fdrtool algorithm.
`statistic`	One of "normal" (default), "correlation", "pvalue". This species the null model.
`verbose`	A logical indicating if status messages should be printed out. Defaults to `TRUE`.
`cutoff.method`	One of "fndr" (default), "pct0", "locfdr".
`pct0`	The fraction of data x_red that will be used for fitting null model - only used if cutoff.method="pct0".

Details

See package fdrtool.

Value

A list with the following components:

x The input vector of the observed test statistics.
x_red The subset of the input vector of the observed test statistics that was given to the fdrtool algorithm.
pval A vector with adjusted p-values for each element of x_red.
pval_full A vector with adjusted p-values for each element of x.
qval A vector with q-values (Fdr) for each element of x_red.
qval_full A vector with q-values (Fdr) for each element of x.
lfdr A vector with local fdr values for each element of x_red.
lfdr_full A vector with local fdr values for each element of x.
statistic The specified type of null model.
param A vector containing the estimated parameters (eta0 (the null proportion of x_red) and the free parameter of the null model).
x0 The cut off value of x_red, indicating the separation between the null model and the mixture model.
f.pval, F.pval, fdr.pval, Fdr.pval, f0, F0, get.pval, fdr, Fdr Functions needed to make diagnostic plots.

References

Strimmer, K. (2008a). A unified approach to false discovery rate estimation. BMC Bioinformatics 9: 303. Available from http://www.biomedcentral.com/1471-2105/9/303/.

Strimmer, K. (2008b). fdrtool: a versatile R package for estimating local and tail area- based false discovery rates. Bioinformatics 24: 1461-1462. Available from http://bioinformatics.oxfordjournals.org/cgi/content/abstract/24/12/1461.