[BioC] DEXSeq, looking for a clever way for interpretation of results
delasa at vt.edu
Tue Jun 11 21:54:13 CEST 2013
Hi Alejandro / Simon
I have done DEXSeq for testing differential exon usage in my time course
RNA-seq data (10 time points).
I have some problems in analyzing/interpreting the results of DEXSeq and I
would be appreciated if you could help me find my answers.
I have tested differential exon usage using DEXSeq in two modes: 1. For all
time course, 2. pairwise comparison of each two consecutive time points.
I have attached one of the genes that from the annotation file, we know
that there are about three isoforms and DEXSeq is showing that during the
time course, there are multiple exons that have been significantly
differentially expressed. In pairwise comparisons, I can see that during
one of the time points, it has gone though significant exon level DE.
I was wondering if there is any way to go from differential exon usage to
isoform regulation and isoform expression in DEXSeq? How can I interpret
the DE results when more than one exon is differentially expressed?
Here in the attached plots, I can see that some exons have significant DE,
but I can not say which transcript (isoform) is effected, or even which one
has been expressed in our experiment at all.
Talking about splicing and isoform regulation can not be done without
knowledge about different isoforms that have bee effected. Notice that I am
not looking for isoform level DE, which I know it is not what DEXSeq does,
but I am looking for the effect of differential expression in exon level
which causes alternative splicing.
I have another problem with interpretation of the plot called "fitted
splicing". If you could explain how you are inferring splicing, I would be
appreciated. Is it somehow related to actual splicing of a gene which we
can observe in the isoforms? I have attached one of the plots to have an
example. Thanks in advance,
Genetics, Bioinformatics, and Computational Biology (GBCB) PhD Student
More information about the Bioconductor