bremt: Bootstrap Resampling Multiple-Testing

In pwirapati/acdx: Aggregated Cell Differential Expression

Description

Calculate various significance testing statistics from a genome-wide scores with bootstrap replicates.

Usage

  bremt(theta, T0=0.2)

Arguments

theta	A matrix of genes (rows) versus bootstraps (columns) containing the score of interest (bigger value is better). The original (non-bootstrap) values are in the first column.
T0	Test statistics value below which null is assumed.

Details

This function computes per-comparison error rate (PCER), family-wise error rate (FWER) and false-discovery rate (FDR) from a set of gene-wise scores with bootstrap replicates.

FWER and FDR are calculated using algorithm in box 2 and 5, respectively, of Ge, Dudoit, Speed (2003) Test 12:1.

Value

A numeric matrix with genes as the row (in the original input order) and columns:

theta	The first column of the original input, typically the log ratios or any functional of the fitted parameters.
SE	Bootstrap standard error, after centering using the original value (not the mean).
T	theta/SE
PCER	Bootstrap one-sided p-value for each gene. The null distribution is from the centered bootstrap replicates.
FWER	Multiple testing p-value based on the step-down MaxT method
FDR	False discovery rate.

Examples

  ## see the 'quick tutorial'

pwirapati/acdx documentation built on Jan. 11, 2021, 12:31 a.m.