Skip to contents

Quantile Normalisation

Source: R/quantile_normalisation.R
quantile_normalisation.Rd

Performs quantile normalisation of a numeric matrix in native R, matching the algorithm used by `preprocessCore` (including its tie-handling rule).

Usage

quantile_normalisation(x)

Arguments

x

A numeric matrix; rows are features (e.g., genes), columns are samples/arrays.

Value

A numeric matrix of the same dimensions as x, quantile normalised.

Details

This function is a native R implementation of the standard quantile normalisation algorithm. It is designed to be a drop-in replacement for, and produce identical results to, the function of the same name in the `preprocessCore` package.

This native R version is provided within `damidBind` to avoid known issues where the `preprocessCore` package can lead to errors or cause R to crash on some Linux systems due to conflicts with OpenMP and/or BLAS/LAPACK library configurations. By providing this native R implementation, `damidBind` ensures it works reliably for all users without requiring them to recompile dependencies or manage system environment variables.

This implementation exactly mirrors the behaviour of the `preprocessCore` library’s classic quantile normalisation, including its specific handling of ties: average ranks are computed for ties, and if the fractional part of a rank is greater than 0.4, the output value is the average of the two adjacent quantile means; otherwise, only the lower (floored) quantile mean is used.

The function stops if any NA, Inf, or NaN values are present in x.

Examples

set.seed(1)
x <- matrix(rnorm(9), nrow = 3)
quantile_normalisation(x)
#>            [,1]       [,2]       [,3]
#> [1,]  0.0929451  0.8390829 -0.3895560
#> [2,]  0.8390829  0.0929451  0.8390829
#> [3,] -0.3895560 -0.3895560  0.0929451