Hi,

I have a question regarding the test of differential isoform usage in IsoformSwitchAnalyzeR. I have a large dataset that includes 28 different pairwise comparisons (I actually only need a subset of these, but it didn't seem like there was a way to control which pairwise comparisons are performed). In some cases I have a very focused a priori hypothesis about differential isoform usage in one specific gene in a few different comparisons. I am wondering if it is possible to get the uncorrected pvalues for these tests rather than the qvalues that are reported with extractTopSwitches? Since I am only focused on one gene, I think the qvalue correction might be overly conservative. In addition, the hypothesis has to do with NMD sensitive transcripts--there are four total for this gene. Would there be any way to combine the data for the four transcripts for the test of differential isoform usage? Our hypothesis is that there will be increased usage of NMD sensitive transcripts overall in one condition, but it doesn't really matter which transcript. Thanks in advance for your help!

Code should be placed in three backticks as shown below

# include your problematic code here with any corresponding output 
# please also include the results of running the following in an R session 

sessionInfo( )

Hi

Sorry for the late response. Had somehow accidentally turned off notifications.

A few answers: 1) you can control which comparisons are made by using the comparisonsToMake argument in importRdata(). 2) If you want to keep raw p-values from the test you have to set reduceToSwitchingGenes=FALSE in isoformSwitchTestDEXSeq(). Then you can find the full info in the "$isoformSwitchAnalysis" entry of the switchAnalyzeRlist as shown here for the example data:

data("exampleSwitchListIntermediary")

head(exampleSwitchListIntermediary$isoformSwitchAnalysis)

3) Although currently not supported by IsoformSwitchAnalyzeR isoform level p-value aggregation have been demonstrated to be beneficial. You can find an example (along with R implementation and recommendations in this article).

Hope this helps.

Cheers Kristoffer