normalize.los: LVS/LOS normalization

In raim/segmenTools: Tools for Functional Genome Exploration

LVS/LOS normalization

Description

This function normalizes the columns of a matrix by a set of invariant data rows (LOS, a "least-oscillating set") to be defined by the user (option los.id). For each column a LOESS fit, local fitting of a 2nd order polynomial with default parameters of the R stats function loess is calculated between the values of the reference set (LOS) and their median or median (option norm.ref over all columns. This is a modified version of the function 'normalize_LVS' from package FLUSH.LVS (Calza et al. 2008), as used for normalization by a "least-oscillating" set of "microarray" probes from a transcriptome time-series of "yeast respiratory oscillations" by Machne & Murray, 2012. Alternatively to LOESS but un-tested, the data can also be normalized by smoothing or simple min/max scaling. Also un-tested and highly experimental: to avoid cutting data at highest values of the LOS set, one can add.maxima of each column to the LOESS fits, or after the fit and normalization reset (reset.extrema) or re-scale (extend.max) maximal values.

Usage

normalize.los(
  x,
  los.id,
  norm.ref = c("median", "mean"),
  norm.type = c("loess", "smooth", "scale"),
  add.maximum = FALSE,
  reset.extrema = FALSE,
  extend.max = FALSE
)

Arguments

x	a data matrix to be normalized
los.id	a numerical or logical vector, providing the indices or a TRUE/FALSE vector for a set of "least-oscillating" or "least-variant" data rows
norm.ref	take the "median" or the "mean" of the reference data set (los.id) for normalization
norm.type	use "loess", "smooth" or "scale" normalization; see Details
add.maximum	untested, see Details
reset.extrema	untested, see Details
extend.max	untested, see Details

References

Calza S, Valentini D, Pawitan Y (2008) Normalization of oligonucleotide arrays based on the least-variant set of genes. BMC Bioinformatics 9:140.

Machne R, Murray DB (2012) The yin and yang of yeast transcription: elements of a global feedback system between metabolism and chromatin. PLoS One 7(6):e37906

raim/segmenTools documentation built on April 25, 2024, 10:14 a.m.