Merge events in SE.MATS.JCEC.txt

[Yin Yu]

Hi,

In the SE event output, there are same skipped exons (same starting and end coordinates, e.g., 504823 and 504888 below), but with different event IDs and different up/downstreamES/EE:

ID

GeneID

geneSymbol

chr

strand

exonStart_0base

exonEnd

upstreamES

upstreamEE

downstreamES

downstreamEE

40705	ENSG00000023191.16	RNH1	chr11	-	504823	504888	502061	502249	506608	506759
40706	ENSG00000023191.16	RNH1	chr11	-	504823	504888	502061	502249	507112	507300

I wonder if I can merge them into One event (because I want to give the exon annotation) , and frankly I am quite confused about the biological meaning of >1 event counts while our focus is to identify the differentially used exons between two conditions.

Thank you!

Best,

Yin

[Thomas Danhorn]

These are different events, because the splice junction at the 5'-end of
the potentially skipped exon is different (in the table it is labeled as
"downstream", but the gene is on the (-)-strand, so it really "upstream").
If you are looking at junction spanning reads only, and you see both
events, then they are both real (check SE.MATS.JC.txt, or even counts in
JC.raw.input.SE.txt are an indication that the junction exists). If only
one or none is in SE.MATS.JC.txt, but you see them both in
SE.MATS.JCEC.txt, you can't be so sure (to figure out what is going on, it
may help to examine the exons and junctions involved in the events in your
BAMs with something like IGV). In that case, if you have indications that
one these events might not be "real" in your case, you can remove the
transcript(s) containing the junction you don't like from the GTF and
rerun rMATS with the modified GTF, but I would only suggest that if you
know what you are doing.

Hope this helps,

Thomas

> --
> You received this message because you are subscribed to the Google Groups "rMATS User Group" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to rmats-user-gro...@googlegroups.com.
> To view this discussion on the web visit https://groups.google.com/d/msgid/rmats-user-group/be39248b-d715-4d59-b9c6-4a93b283d97fo%40googlegroups.com.
>

[Yin]

Hi Thomas,

My purpose is to annotate All the Exons in the SE output from rMATS. For example, as I posted the question previously, one way is to transfer SE output to BED format, then to nucleotide sequence and protein sequence. 

Although these are two different events,  in the BED format, given they have same start (second column in BED) and end (third column in BED) coordinates, they will have only One nucleotide sequence, regarding this one skipped exon.

Well since there could be different outputs between SE.MATS.JC.txt and SE.MATS.JCEC.txt, which one would you suggest to use?

Thank you Thomas!

Best regards,

Yin

[Thomas Danhorn]

Hi Yin,

The difference between SE.MATS.JC.txt and SE.MATS.JCEC.txt is that the
former *only* uses junction-spanning reads, whereas the latter adds reads
don't span junctions, but are indicative of one transcript over another
(e.g. reads falling inside the potentially skipped exon). SE.MATS.JC.txt
is more stringent, as a read spanning a particular junction is a very
strong indication that this junction actually exists. SE.MATS.JCEC.txt is
less stringent, but more sensitive, as with low read coverage the number
of junction-spanning reads become be too low as a reliable input for the
statistics, resulting in higher p-values if they are used as the only
reads.

So you which one you use depends -- do you want have few false-positives
at the cost of more false-negatives (use SE.MATS.JC.txt), or do you want
to keep false-negatives low, as the cost of more false-positives (use
SE.MATS.JCEC.txt). How you answer this may depend on the number of events
you get overall (if there are a lot, you can afford to lose some to make
the reamining ones more accurate).

As for your skipped-exon annotation project, you have to ask yourself if
you really only care about the skipped exon, or also about its context.
As your example shows, the same exon can be skipped through two (or more)
different splicing events, each of which may be controlled by different
factors. Think about what this means for your big, overarching question.

Hope this helps,

Thomas

[tian wu]

Hello, there are some problems here. For me to run the same with rMATS, my result file ses.mats.jce.txt is empty. Do you have any suggestions?

[kutsc...@gmail.com]

See these issues for possible reasons for empty output:
https://github.com/Xinglab/rmats-turbo/issues/89
https://github.com/Xinglab/rmats-turbo/issues/95

Eric