Compute expression variability measure

Share:

Description

This function computes expression variability in a way that removes dependence on mean expression. It uses a local polynomial likelihood method to estimate variance as gamma distributed around given mean expression for each probeset. This function makes this calculation using all samples in argument. To calculate expression variability for samples in different groups, call this function for each subset of columns separately.

Usage

1
ev(x, cutoff = NULL, plot = FALSE, ...)

Arguments

x

matrix of gene expression, with one column per sample

cutoff

minimum expression value to be included in computation (for frma normalized data, we find 2.54 to be a good value for determining if a probeset is expressed in a given sample (default NULL)

plot

make a plot of local likelihood model using smoothScatter (default=FALSE)

...

arguments passed to smoothScatter

Value

numeric vector of length equal to number of rows of x

Author(s)

Hector Corrada Bravo hcorrada@gmail.com

References

E. Alemu, H. Corrada Bravo, S. Hannenhalli (2014). Determinants of Expression Variability. Nucleic Acids Research, 42 (6), 3503-14.

See Also

frma for normalization

Examples

1
2
3
4
5
if (require(antiProfilesData)) {
  data(apColonData)
  e <- exprs(apColonData)[,pData(apColonData)$Status==1]
  ev <- ev(e, cutoff=2.54)
}