Dear DEXSeq developers,
I have a question regarding running DEXSeq with more than 2 conditions.
I have expression data for 4 different conditions-temperatures, each of them in 3 replicates.
Is there a possibility that I run first 4 steps of DEXseq for all temperatures so the dispersion estimates are more correct, i.e. so the normalization captures sampling variance between all the temperatures like this:
and then in the testForDEU step I choose for the function to report only p-values for a certain pairwise comparison between temperatures (and not the p-value which tells me if there was at least one temperature with significant effect, but doesn't tell me which one), something like this:
I hope my question is clear enough and not to crazy
All the best,
Ana Marija
I have a question regarding running DEXSeq with more than 2 conditions.
I have expression data for 4 different conditions-temperatures, each of them in 3 replicates.
Is there a possibility that I run first 4 steps of DEXseq for all temperatures so the dispersion estimates are more correct, i.e. so the normalization captures sampling variance between all the temperatures like this:
Code:
ecs=read.HTSeqCounts(countfiles=countfiles[c(1:3,4:6, 7:9, 10:12)],design=factor(c("T13","T13","T13","T18","T18","T18","T23","T23","T23","T29","T29","T29")),"/Volumes/Temp/Anna/counting/dmel-all-transcript-r5.49-oregonR-forDEXseq-notaggr.gff") ecs=estimateSizeFactors(ecs) ecs=estimateDispersions(ecs) ecs=fitDispersionFunction(ecs)
Code:
ecs1318=testForDEU(ecs, comparison="T13","T18") ##made up option :) ecs1318=estimatelog2FoldChanges(ecs1318) ecs1323=testForDEU(ecs, comparison="T13","T23") ecs1323=estimatelog2FoldChanges(ecs1323) #and so on...
All the best,
Ana Marija
Comment