Does anyone know how to contrast only a single pair of conditions in JunctionSeq (which is based on DEXseq AFAIK), if there are > 2 levels of the condition? I would like to test the null hypothesis that there is no condition:bin effect between one pair of conditions, but still using information from the untested conditions to better estimate the dispersion.
For example, in the JunctionSeq manual pg 40 they give code for testing a single, 3-level condition in the example data:
threeLevelVariable <- c("GroupA","GroupA", "GroupB","GroupB", "GroupC","GroupC"); jscs <- runJunctionSeqAnalyses(sample.files = countFiles, sample.names = decoder$sample.ID, condition=factor(threeLevelVariable), flat.gff.file = gff.file, nCores = 1, analysis.type = "junctionsAndExons" );
How could I request a Group A vs Group B contrast, without throwing away the information on dispersion contained in the data for Group C?