duffyRMA: Robust Multichip Average (RMA)

In robertdouglasmorrison/DuffyTools: Duffy Lab Utility Tools

Robust Multichip Average (RMA)

Description

Our local implementation of RMA normalization

Usage

duffyRMA(m, targetBGlevel = NULL, magnitudeScale = NULL, columnSets = list(all = 1:ncol(m)))

duffyMagnitudeNormalize(m, globalSum = 1000 * nrow(m))
duffyRMA.bgSubtract(m, targetBGlevel = NULL)
duffyRMA.qn(m, columnSets = list(all = 1:ncol(m)))

Arguments

m	numeric matrix of expression data, with one row per oligo/gene and one column per sample
targetBGlevel	background correction target value. NULL means use 'DensityKernelMode'
magnitudeScale	mean gene intensity target value (aka Quackenbush). NULL means no global scaling step
columnSets	a list of vectors of column numbers, to do RMA on subsets of columns of m
globalSum	total sum of gene intensity target value (aka Quackenbush)

Details

This implementation of RMA gives more layers of flexibility. Global scaling in the style of Quackenbush (2002), with settable average gene intensity, is the first step. Next, the adjustment for background hybridization levels can be set to an explicit value, or calculated from the mode of the density kernel. Lastly, the quantile normalization step is run. Each of these functions is callable separately for finer control.

Value

a matrix of the same size as m, with normalized expression values

Author(s)

Bob Morrison

References

Irizarry, B. & Bolstad, B. (2003) Nucleic Acids Research 31(4):e15

See Also

getGlobalMetrics, for various metrics of an expression matrix. DKMshift, for details on applying the density kernel mode background correction. rankEquivalentIntensity, for simple quantile normalization.

robertdouglasmorrison/DuffyTools documentation built on April 16, 2024, 6:31 a.m.