normRepLoess: Bootstrap of LOWESS normalisation

Description

This function normalises a microarray object re-doing the LOWESS fitting several times, selecting a pre-specified proportion of points aleatorily.

Usage

normRepLoess(raw, span=0.4, propLoess=0.5, nRep=50, func="none", bkgSub="none", ...)

Arguments

raw

an object of class maigesRaw to be normalised.

span

real number in (0,1) representing the proportion of points to use in the loess regression.

propLoess

real number in (0,1) representing the proportion of points (spots) to be used in each iteration of loess.

nRep

number of repetitions for loess procedure.

func

character string giving the function to estimate the final W value. You must use 'mean', 'median' or 'none' (default).

bkgSub

character with background subtraction method, using the function backgroundcorrect from limma package.

...

additional parameters for function loessFit from limma package.

Value

The result of this function is an object of class maiges.

Details

The LOWESS fitting for normalising microarray data is a computational intensive task, so pay attention to not specify a very large argument in nRep. If you do so, your process will take so much time to conclude.

Examples

Run this code

## Loading the dataset
data(gastro)

## Doing the repetition loess with default parameters. Be carefull, this
## is very time consuming
## Not run: 
# gastro.norm = normRepLoess(gastro.raw2)
# ## End(Not run)

## Do the same normalization selecting 60% dos spots with 10
## repetitions and estimating the W by the mean value.
## Not run: 
# gastro.norm = normRepLoess(gastro.raw2, propLoess=0.6, nRep=10, func="mean")
# ## End(Not run)