I am using the rlog transformation from DEseq2 (see first figure) and it produces a strange graph. 1th graph are random numbers. 2nd are the raw counts. 3rd are normalized raw counts etc.
Question1): In the 5th graph- Why are the red dots so close to the x axis? It should more or less look like the first graph. Any ideas on whats causing this?
After some thinking I went on and reduced my data keeping only genes that are expressed in among ALL samples (second figure).
Here rlog does what its supposed to do.
Question2) While we are at it- Is this the correct order for normalization (estimateSizeFactors) + VST(variance stabilizing transformation):
It cant be anything else since rlogTransformation(dds) only takes a DEseqdataset as input with integers. After estimateSizeFactors(dds) and obtaining the counts with counts() the integers are transformed to double. So there is no way I could use rlogTransformation(counts(dds,normalized=T)), right?
Thanks a lot!
For the sake of completeness- the figures are generated with the "vsn" package and the transformations with "DEseq2".