dexseq simulation

[Pedrissimo]

Hello,

I am comparing two drug treated samples to two control samples, both are biological replicates. I'd like to test how different levels of dispersion effect my samples by varying the level of dispersion for each sample. How can I assign different artificial dispersion levels to each pair of biological replicates?

[areyes]

dispersion columns in DEXSeq

Hi,

I think I did not understand exactly what you want to do with your simulations, but the way DEXSeq stores the dispersion parameters is the following (demonstrated in the pasilla data package):

data("pasillaExons", package="pasilla")
pasillaExons <- fitDispersionFunction(estimateDispersions( estimateSizeFactors( pasillaExons ) ))
fData(pasillaExons)$dispBeforeSharing
pasillaExons@dispFitCoefs
fData(pasillaExons)$dispFitted
fData(pasillaExons)$dispersion

after calling estimateDispersions and fitDispersionFunction, dispBeforeSharing will contain the Cox-Reid dispersion estimates, dispFitted will contain the mean-variance fitted values for the exons respective count value (pasillaExons@dispFitCoefs contains the information from the function itself), and the dispersion column will contain the maximum between the two previous and this last is the one use for testing. I guess you could modify this last column for your simulations... but, what exactly you want to see when you simulate this dispersion values?