R: Estimate dispersions

estimateDispersions {DEXSeq}

R Documentation

Estimate dispersions

Description

This function estimates for each counting bin of the ExonCountSet object a dispersion value. It stores these values in fData(ecs)$dispBeforeSharing.

Usage

## S4 method for signature 'ExonCountSet'
estimateDispersions( object,
         formula = ~ sample + exon + condition : exon,
         minCount = 10, nCores = 1 )

Arguments

`object`	An ExonCountSet object.
`formula`	Formula used in the GLM to estimate the dispersion values. The terms in the formula must be design columns of the ExonCountSet object, the l.h.s. will be the counts for each exon.
`minCount`	Counting bins with less than `minCount` counts (summed over all samples) are skipped in the tests. This reduces computation time, as counting bins with very few counts cannot give a significant signal anyway. For skipped counting bins, the `testable` column in `fData` is set to `FALSE`.
`nCores`	Number of cores to be used to estimate the dispersions. The `parallel` package must be loaded in order to spread the job onto several cores.

Details

For the dispersion estimation, we use the Cox-Reid conditional maximum likelihood method of McCarthy et al. (Nucl Acid Res., 2012, 40:4288), which they devised for the edgeR package.

Value

An object of class ExonCountSet with dispersion featureData(object)$dispersion_CR_est) parameters filled).

Examples

if(suppressWarnings(require("pasilla", quietly=TRUE, character.only=TRUE))){

   data("pasillaExons", package="pasilla")
   pasillaExons <- estimateSizeFactors( pasillaExons )
   pasillaExons <- estimateDispersions( pasillaExons )
   head( fData(pasillaExons)$dispBeforeSharing )

}

[Package DEXSeq version 1.8.0 Index]