View Single Post
Old 12-12-2012, 12:49 PM   #8
zimmernv
Junior Member
 
Location: Dallas, Texas

Join Date: Dec 2012
Posts: 4
Default thanks ... one more question

Hey Alejandro,

Thanks for the response. That fixed the issue and I can now visualize my results.

Somewhat astonishingly, the differential exon usage test does not yield any positive differential splicing results for my dataset (2 replicates for both the controls and the transgenics, with very high count numbers). For a whole-transcriptome 45 million reads (i.e. decent coverage) RNA-seq experiment, not seeing a single alternative splicing event is a bit odd, don't you think?

One area of concern is the use of the non-overlapping exonic regions made via the dexseq_prepare_annotation.py script. I can understand the motivation for creating these regions, but I'm struggling to see how I can then convert the final output (counts assigned to new but non-biological exonic regions) to biologically-meaningful outputs (i.e. deconvolving the counts over these exonic regions into counts over the exons that helped to build the exonic region). To take an example, the Apoer2 gene in M. musculus has 20 exons. But the dexseq script produces a gff file that lists Apoer2 as having 40 exonic regions due to the overlaps of transcripts. The DEXSeq plots then show 40 exons rather than the conventional 20. How can I get an image that would reveal the counts over the 20 exons? Should I manually change the exonCountSet to reflect the biological reality?

To summarize the difficulty I'm having: DEXSeq, as I see it, is testing for Differential Exonic Regions rather than Differential Exons (where only the latter is biologically-meaningful, or so I think ... I come from the computer science side more so than biology).

I eagerly await your critique and insights.
zimmernv is offline   Reply With Quote