rMATS skipping 2 exons that are expressed

[mlcsmits]

Hey, rMATS team, I was running rMATS and rmats2sashimiplot, and this generated the output you see below: It says there are no junction-spanning reads between exons 2 and 3, so these exons are skipped. However, there are visible reads on these exons. I tried to run it with an anchor length of 24 because the first 8 bases of exons 2 and 4 are the same, but still, I got the same output sashimi plot. Could you please explain what could be wrong or what I don't understand and maybe how to fix it?

[EricKutschera]

It looks like that plot is from rmats2sashimiplot using the -e EVENTS_FILE argument. With -e only the junctions for that event are shown. My guess is that there are junction reads between some of the other exons, but they're just not shown. If you run rmats2sashimiplot with -c instead of -e then it will show the counts for any junction in the gff3 file for those coordinates

[mlcsmits]

Thanks for your quick response; the solution worked. There is no junction between exon 2 and 3 and only 15 between exon 3 and 4, which is very low. But between exon 1 and 2, there are around 12,000 reads, so rMATS do not notice that. What could cause that?

[EricKutschera]

I'm not sure what you mean by rMATS not noticing the junction between exon 1 and 2. The initial sashimiplot you posted was for a specific SE event and so only the 3 junctions for that event had counts shown in the plot. From the other sashimiplot it doesn't look like there is an SE event where exon 2 is skipped. It doesn't look like there are any reads to support a junction from exon 2 to any other exon

[mlcsmits]

Thank you for your explanation, I will discuss it with my supervisor.