Median-balanced quantile normalization

mbqnGetNRIfeatures

R Documentation

Identify rank invariant (RI) and nearly rank invariant (NRI) features

Description

Compute the rank frequency of each feature of a matrix and identify NRI/RI features.

Usage

mbqnGetNRIfeatures(x, low_thr = 0.5, method = NULL, verbose = TRUE)

Arguments

`x`	a data matrix. Rows represent features, e.g. protein abundances; columns represent samples.
`low_thr`	a value between [0 1]. Features with RI frequency >=`low_thr` are considered as NRI/RI; default 0.5.
`method`	character specifying function for computation of quantile normalization; "limma" (default) for `normalizeQuantiles()` from the limma package or "preprocessCore" for `normalize.quantiles()` from the preprocessCore package.
`verbose`	logical indicating to print messages.

Details

Quantile normalize the data matrix and sort ranks. Determine the maximum frequency of equal rank across all columns for each feature. Features with maximum frequency above the user-defined threhold are declared as nearly rank invariant.

Value

A list with elements:

`p`	a matrix with the rank invariance frequencies `ri.freq` and the sample coverage `sample.coverage` for all detected RI/NRI features
`max_p`	maximum rank invariance frequency in percent
`ip`	index of feature with maximum rank invariance frequency
`nri`	table of the rank invariance frequencies in percent for each NRI/RI feature
`var0_feature`	indices of features with zero sample variance after QN
`low_thr`	threshold used for NRI/RI detection from RI frequency.

Author(s)

Ariane Schad

References

Brombacher, E., Schad, A., Kreutz, C. (2020). Tail-Robust Quantile Normalization. BioRxiv.

Examples

## Check data matrix for RI and NRI features
set.seed(1234)
x <- mbqnSimuData("omics.dep")
RI <- mbqnGetNRIfeatures(x, low_thr = 0.5, verbose = FALSE)
mbqnPlotRI(RI)

arianeschad/MBQN documentation built on April 1, 2022, 7:08 a.m.