doSVA: Perform surrogate variable analysis (SVA) to an EdgeObject...

In bedapub/ribiosNGS: Utilities for next-generation sequencing data analysis in ribios

Perform surrogate variable analysis (SVA) to an EdgeObject object

Description

Perform surrogate variable analysis (SVA) to an EdgeObject object

Usage

doSVA(edgeObj, transform = c("voom", "cpm"))

Arguments

edgeObj	An EdgeObject object
transform	Function name to perform transformation, currently supported values include voom and cpm The count data associated with the EdgeObject object is first transformed, and surrogate variables are estimated from the transformed data. Correspondingly the design matrix and contrast matrix associated with the object are updated, too.

Examples

set.seed(1887)
exMat <- matrix(rpois(12000, 10), nrow=2000, ncol=6)
exMat[1:100,2:3] <- exMat[1:100, 2:3]+20
exGroups <- gl(2,3, labels=c("Group1", "Group2"))
exDesign <- model.matrix(~exGroups)
exContrast <- matrix(c(-1,1), ncol=1, 
              dimnames=list(c("Group1", "Group2"), c("Group2.vs.Group1")))
exDescon <- DesignContrast(exDesign, exContrast, groups=exGroups)
exFdata <- data.frame(GeneSymbol=sprintf("Gene%d", 1:nrow(exMat)))
exPdata <- data.frame(Name=sprintf("Sample%d", 1:ncol(exMat)),
                     Group=exGroups)
exObj <- EdgeObject(exMat, exDescon, 
                     fData=exFdata, pData=exPdata)
exSVAobj <- doSVA(exObj, transform="voom")
designMatrix(exSVAobj)
contrastMatrix(exSVAobj)

## Note that the SVA is sensitive against parameterisation, see 
## the example below. Also notice that in the zero-intercept parameterisation, 
## the SVA does not give meaningful results.
designMatrix(exObj) <- model.matrix(~0+exGroups)
designMatrix(doSVA(exObj, transform="voom"))

bedapub/ribiosNGS documentation built on Feb. 10, 2025, 12:34 a.m.