edgeRun (version 1.0.8)

edgeRun-package: Unconditional exact testing for two-group digital expression data

Description

This package provides functions for testing for differential expression for digital data (e.g. RNA-Seq, CAGE-Seq, ChIP-Seq, etc.). Exact unconditional tests based on the negative binomial distribution are provided. Similar functionality exists in the Bioconductor package edgeR where a similar but conditional exact test is implemented. The unconditional test is more powerful, especially for lowly expressed genes or when sample size is extremely small.

Details

ll{ Package: edgeRun Type: Package Version: 1.0.03 Date: 2014-04-09 License: MIT } Users familiar with edgeR can use edgeRun functionality by simply using the UCexactTest function instead of edgeR's exactTest function. We recommend the edgeR workflow that takes input count data and ends up with a DGEList. Please refer to edgeR documentation on how to use that workflow. NOTE: edgeRun can take several hours to run due to computational complexity, and is only recommended for experiments with very few samples where the power benefit of edgeRun is more evident. Power gains for larger sample sizes might not be worth the extra running time, in those cases edgeR is more appropriate and much faster.

References

Dimont, E., et al. edgeRun: an R package for sensitive, functionally relevant differential expression discovery using an unconditional exact test. bioRxiv doi: http://dx.doi.org/10.1101/008409 Robinson, MD, et al. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 2010;26:139-140.

See Also

exactTest,DGEList

Examples

# Example taken from edgeR documentation:
# generate raw counts from NB, create list object
# this creates 50 genes across 4 samples, 2 from each group
y <- matrix(rnbinom(50*4,size=1/0.2,mu=10),nrow=50,ncol=4)
d <- DGEList(counts=y, group=c(1,1,2,2), lib.size=colSums(y))
d <- calcNormFactors(d)
d <- estimateCommonDisp(d)
d <- estimateTagwiseDisp(d)

#using edgeR CONDITIONAL exact test
de.edgeR <- exactTest(d) 
topTags(de.edgeR)

#using edgeRun, UNCONDITIONAL exact test
#argument 'upper' specifies the number of iterations
#higher values give more accurate p-values but take longer to run
#see Supplementary Methods (Dimont, et al. 2014) for details
de.edgeRun <- UCexactTest(d,upper=50000) 
topTags(de.edgeRun)


# same p-values using low-level function directly (edgeR)
p.value.edgeR <- exactTestDoubleTail(y[,1:2], y[,3:4]
,dispersion=d$tagwise.dispersion)

p.value.edgeRun <- pvalue(s1=rowSums(y[,1:2,drop=FALSE])
, s2=rowSums(y[,3:4,drop=FALSE]), phi=d$tagwise.dispersion, n1=2, n2=2)