rMATS not detecting alternative event

[Matt Morgan]

Hi, 

I am attempting to detect an alternative event that you can see here in IGV using rMATS in the screenshot. 

There is a slight adjustment in the 5'SS of the preceding exon (- strand), which results in the inclusion of the cassette exon in yellow. 

However, rMATS does not appear to be picking the event up at all, and it is not present in any of the fromGTF files - this is despite the alternative event being annotated in the GTF and other alternative events from the same transcript being picked up. 

I have tried adding in the event manually to the fromGTF files...I have a single line in my fromGTF.SE.txt file....

ID      GeneID  geneSymbol      chr     strand  exonStart_0base exonEnd upstreamES      upstreamEE      downstreamES    downstreamEE

1       "ENSG00000188529"       "SRSF10"      chr1    -       23977910        23978273        23974974        23975051         23978653        23978817

However, this does not appear to be working... when I run the following the event deletes itself in the fromGTF files and all output files are empty. 
#adjust paired/unpaired and readLength as necessary
rmats.py \
--b1 "/mnt/research2/mjm280/${project_name}/extract_bam/norm.txt" \
--b2 "/mnt/research2/mjm280/${project_name}/extract_bam/tum.txt" \
-t single \
--gtf "$ref_gtf" \
--readLength 75 \
--variable-read-length \
--nthread 25 \
--od "$out_dir" \
--tmp "$tmp_dir" \
--paired-stats \
--fixed-event-set "/mnt/research2/mjm280/${project_name}/testrmats" \
--task both \
--statoff

I would understand if the rMATS quantification might look different to how the IGV visualisation looks, but why is no event being produced at all? Is it because it is a combination of 2 alternative events? 

I would love some pointers for this if possible!  

Thank you!

[kutsc...@gmail.com]

This post has some details about how events are detected and also filtered: https://github.com/Xinglab/rmats-turbo/issues/467#issuecomment-2605445497

In this case the event doesn't fit any of the event diagrams from the README. The highlighted exon doesn't fit the SE definition since the "skipping" junction doesn't use the same upstream splice site as the inclusion junction. It also doesn't fit the A5SS definition since the alternative splice site is always used with the highlighted exon, but the other splice site is always used with the downstream exon. rMATS can't detect it because it doesn't fit any of the event types

I can't tell the coordinates from the screenshot, but the line you added to the fromGTF.SE.txt file may have used the alternative splice site where reads support the junction to the highlighted exon. In that case the event would have no reads supporting the skipping isoform. By default that would lead to the event being filtered out, however you have --statoff in your command which should disable the filter. With both --fixed-event-set and --statoff in the command I would expect all events defined in --fixed-event-set to be in the final --od

Did you use the same directory for both --od and --fixed-event-set? I think that may cause an issue. If there's still no output when a separate --fixed-event-set is used can you post the output that rmats.py prints? Hopefully there is some error message

Eric