Question

DESeq2 on CIRI results ?

0

Entering edit mode

sbcn ▴ 80

@sbcn-4752

Last seen 2.3 years ago

Spain

Hi,

I am analysing a RNA-seq data set of 6 samples (2 conditions in biological triplicates); the goal is to identify the differential expressed circular RNAs in one condition versus the other.

I used the CIRI2 program (v2.0.6) to detect circRNAs from the RNA-seq data.

As an output, I get the circular junction read counts of predicted circRNAs (back-spliced junction read) and now I was wondering if using DESeq2 for analysing the differential expression of those predicted circRNAs was a good approach, based on those counts?

Alternatively, could anyone advise a better approach?

Thank you!

Sarah

deseq2 circrna ciri ciri2 • 1.9k views

ADD COMMENT • link 6.6 years ago sbcn ▴ 80

0

Entering edit mode

Hi Michael,

Thanks for your reply!

I have counts for around 1000 circRNAs: do you think that is enough?

Thanks again,

Sarah

ADD REPLY • link 6.6 years ago sbcn ▴ 80

1

Entering edit mode

It's not just a question of number, but whether they are all changing or not. I just don't know enough about the biology of your system and circRNAs to assume that at least some of these 1000 are not changing across condition. If you think they aren't changing you could just use these counts I suppose, but otherwise, you could add them to the gene count matrix I think.

ADD REPLY • link 6.6 years ago Michael Love 43k

0

Entering edit mode

I edited my answer to include "at least some"...

ADD REPLY • link 6.6 years ago Michael Love 43k

0

Entering edit mode

Ok I see... Then I'll try and assess the variability in expression across all samples, and see then how I proceed.

Thanks for the support!

ADD REPLY • link 6.5 years ago sbcn ▴ 80

score 1 · Answer 1 · 2018-05-08

How many circRNAs do you have counts for? The critical issue here is that DESeq2 (and all other similar methods) need to compute a library size correction. Without any additional information, it is assumed that there are some rows of the count matrix where there are not large biological differences. It may make sense to add the circRNA counts to a standard gene expression matrix, so you can use that for library size correction.