reb (version 1.50.0)

smoothByRegion: reb

Description

This function “smooths” gene expression data to assist in the identification of regional expression biases.

Usage

reb(eset, genome, chrom = "ALL", ref = NULL, center = FALSE,
aggrfun=absMax, method = c("movbin", "supsmu", "lowess","movt"), ...)

Arguments

eset

the expression set to analyze

genome

an associated chromLoc annotation object

chrom

a character vector specifying the chromosomes to analyze

ref

a vector containing the index of reference samples from which to make comparisons. Defaults to NULL (internally referenced samples

center

boolean - re-center gene expression matrix columns. Helpful if ref is used

aggrfun

a function to summarizes/aggregates gene expression values that map to the same locations. Defaults to the maximum absolute value absMax. If NULL, all values are included.

method

smoothing function to use - either "supmu", "lowess", "movbin" or "movt".

...

additional paramaters to pass along to the smoothing function

Value

An exprSet

Details

reb returns an eset that contains predictions of regional expression bias using data smoothing approachs. The exprSet is separated into subsets based on the genome chromLocation object and the gene expression data within the subsets is organized by genomic location and smoothed. In addition, the approx function is used to estimate data between any missing values. This was implimented so the function follows the ‘principles of least astonishment’.

Smoothing approachs are most straightforwardly applied by comparing a set of test samples to a set of control samples. For single color experiments, the control samples can be specified using the ref argument and the comparisons are generated internal to the reb function. This argument can also be used for two-color experiments provided both the test and control samples were run against a common reference. If multiple clones map to the same genomic locus the aggrfun argument can be used to summarize the overlapping expression values to a single summarized value. This is can be helpful in two situtations. First, the supsum and lowess smoothing functions do not allow for duplicate values. Currently, if duplicate values are found and these smoothing functions are used, the duplicate values are simply discard. Second, if 50 copies of the actin gene are present on a the array and actin changes expression under a given condition, it may appear as though a regional expression bias exists as 50 values within a region change expression. Summarizing the 50 expression values to a single value can partially correct for this effect. The idiogram package can be used to plot the regional expression bias.

References

Furge KA, Dykema KJ, Ho C, Chen X. Comparison of array-based comparative genomic hybridization with gene expression-based regional expression biases to identify genetic abnormalities in hepatocellular carcinoma. BMC Genomics. 2005 May 9;6(1):67. PMID: 1588246

MCR eset data was obtained with permission. See PMID: 15377468

See Also

movbin,idiogram

Examples

# The mcr.eset is a two-color gene expression exprSet
# with cytogenetically complex (MCR) and normal 
# control (MNC) samples which are a pooled-cell line reference.


data("mcr.eset")
data(idiogramExample)

## Create a vector with the index of normal samples
norms <- grep("MNC",colnames(mcr.eset@exprs))

## Smooth the data using the default 'movbin' method,
## with the normal samples as reference

cset <- reb(mcr.eset@exprs,vai.chr,ref=norms,center=TRUE)

## Display the results with midiogram
midiogram(cset@exprs[,-norms],vai.chr,method="i",dlim=c(-5,5),col=.rwb)