dexseq doesn't identify exons as differential
Entering edit mode
Assa Yeroslaviz ★ 1.5k
Last seen 14 days ago
Munich, Germany


I was wondering how dexseq is calculating the differential exon usage. I am having troubles interpreting the results, which I think are significant, but dexseq does not :-)

I have here one of the genes from my run, which looks to me, that it is very much differentially used. the three plots are the results for the counts, the expression and the splicing events calculated by DEXSeq. 

I don't understand the results. To my eyes, it looks as if the counts of exons 26-60 should be differentially expressed, but somehow they're not. On the other hand, the two exon defined as different are exons 17 and 19, which doesn't look at all changed.





splicing: here it looks completely different than the other two plots.


Is there an explanation to that?



dexseq differential exon usage • 1.1k views
Entering edit mode
mat.lesche ▴ 80
Last seen 2.6 years ago

Hi Assa,

DEXSeq doesn't look for differential expression. The tool is designed to look for a differential usage of single exon bins compared to the usage of all other exon bins in a gene. In your case the gene shows an over expression in EB2 compared to the other condition. Hence the high expression in EB2 is normal, except for the two exon bins 17 and 19. Here expression/usage is not on the same level as compared to the other exons. That's why they are marked as significant. The concept of usage is completely different to expression and takes a while to wrap one's head around it. Actually here, DEXSeq Usage and Expression Confusion is a very good example, explained by Simon Anders. That should make things quite clear



Entering edit mode

Hi thanks for the explanation and the link. I think I have understood what the goal of DEXSeq, but than again it still doesn't do what you say, at least not for all the exons. In the gene I have shown above there are almost two separate parts to the gene. (From the biological side, there is an alternative promotor in the region of exon (bin) 26. This is why the change of read counts is so clear. )

But what I don;t understand about the assignment of significant exon bins is that there are also other exon bins with similar behavior before exon 26. So why these two exons? Even these two are not really similar in their expression behavior or count pattern.

Is there a reason why these two suppose to be significant, while other not?


Login before adding your answer.

Traffic: 386 users visited in the last hour
Help About
Access RSS

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6