questions about rMATS output

Question

[zhaopeihua0520]

Hi,

I am new to use rMATS for alternative splicing detection. I have a few questions:

1. i noticed that the default is no detection of novel splice sites (-novelSS 0). Why are there still files like 'fromGTF.novelEvents.A3SS.txt' in the ASEvents folder?

2. AS_Event.MATS.JunctionCountOnly and AS_Event.MATS.ReadsOnTargetAndJunctionCounts generated different results. What are the advantages and disadvantages for these two types of files?

Thanks in advance.

Answer 1

[Charles Warden]

I see that you haven't gotten a response yet for your most in the rMATS discussion group:

https://groups.google.com/forum/#!topic/rmats-user-group/s8-VOy4HFJM

MATS provides splicing events defined from a number of different resources. According to this post, rMATS can define some types of novel junctions:

https://groups.google.com/forum/#!searchin/rmats-user-group/novel%7Csort:relevance/rmats-user-group/8qyvVcalm3I/S_eewwx9BgAJ

Ideally, I would say being able to incorporate additional information would be better for making more accurate predictions. However, if I remember correctly, you'll tend to get more events called from "JunctionCountOnly".

If you have replicates and are willing to try other programs, I like JunctionSeq (basically an extension of DEXSeq) a little better. It only discovers novel junctions between known exons, but I think that is similar to rMATs. It also doesn't provide specific annotations (just differences in coverage for exons and junctions), but I would say it does a decent job of identifying genes with some sort of splicing change and it provides a nice visualization for those differences across your replicates.

http://hartleys.github.io/JunctionSeq/

[zhaopeihua0520]

Thanks for your help.

I am trying to use JunctionSeq. But i don't understand why JunctionSeq tests for differential usage of both exons and splice junctions simultaniously. I think that differential exon usage also means differential splce junction usage.

[Charles Warden]

1) If the splice junction has a change in coverage for corresponding exon(s), that provides additional evidence that the splicing event is real (and vice versa).

2) If there is differential splicing with a novel exon, you would see a change in splicing junction coverage that might not show a clear change among known exons.

If you want to find potential novel exons associated with difference in junction coverage, you would try a program like SGSeq (although I haven't had a chance to do very careful testing of that program):

https://bioconductor.org/packages/release/bioc/html/SGSeq.html