DEXseq: different usage for the shared exon

[capricy]

I read the DEXseq paper (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3460195/) and found that the examples of the differentially used exons (chimpanzee data ) are all existing/missing in some isoforms.

My results revealed some differentially exons are shared in all the isoforms. How should I interpret this kind results?

An example is shown in the attached cartoon. That is, two isoforms: "a" and "b", exist for this gene. The red exon was the differentially used exon based on the calculated p-value between three tissues, brain,intestine, and liver.

Another question about the output: what does it mean by "normalized accounts" and "fitted expression"? Normalized against what?

I have a gene with final baseMean of 73, and in the plotDEXseq graph, the corresponding values of the fitted expression were 40, 50, and over 100. How do those values relate to the baseMean?

Thank you very much for any inputs.

[areyes]

Hi capricy (and sorry for the late reply):

The counts are normalized for sequencing depth. The "fitted splicing", are the fitted values from a generalised linear model corresponding to "exon usage levels". So, imagine you have a gene that is differentially expressed and it has also exons with differential inclusion. The "fitted splicing" will remove those effects caused by the difference in expression levels, and only keep those that are due to differences in exon usage. The mean base is simply the mean of the counts across all the samples.

Best
Alejandro