combineHarmonic: Combine p-values using weights

In pmoulos/metaseqR2-local: An R package for the analysis and result reporting of RNA-Seq data by combining multiple statistical algorithms

Combine p-values using weights

Description

This function combines p-values from the various statistical tests supported by metaseqR using p-value weights.

Usage

    combineHarmonic(p, w, zerofix = NULL)

Arguments

p	a p-value matrix (rows are genes, columns are statistical tests).
w	a weights vector, must sum to 1.
zerofix	NULL (default) or a fixed numeric value between 0 and 1.

Details

The argument zerofix is used to correct for the case of a p-value which is equal to 0 as a result of internal numerical and approximation procedures. When NULL, random numbers greater than 0 and less than or equal to 0.5 are used to multiply the offending p-values with the lowest provided non-zero p-value, maintaining thus a virtual order of significance, avoiding having the same p-values for two tests and assuming that all zero p-values represent extreme statistical significance. When a numeric between 0 and 1, this number is used for the above multiplication instead.

Value

A vector of combined p-values.

Author(s)

Panagiotis Moulos

Examples

p <- matrix(runif(300),100,3)
pc <- combineHarmonic(p,w=c(0.2,0.5,0.3))

pmoulos/metaseqR2-local documentation built on May 21, 2024, 3:46 a.m.