Sebastian Uhrig

Yes, it's 100% because of the low mapping rate.

This is because Arriba ignores all fusion candidates that are explained by normal splicing. If you remove transcripts from the annotation, then Arriba may be misled into thinking that reads...

Yes, there are many transcripts with introns >10k. You can increase the value using the parameter `-R`. Beware that large values run the risk of missing fusions from focal deletions....

> How does Arriba distinguish whether a splicing event is a normal molecular mechanism or not? Every splice junction mentioned in the GTF file is considered normal. > How does...

Your understanding is correct. While they are listed in the column `read_identifiers` for both fusions, they are counted for only one of the fusions in the columns `split_reads1/2` and `discordant_mates`....

This is not possible, unfortunately. However, I don't think this is meaningful either. If a fusion made it to the main output file of Arriba, you should consider all reads...

> I wonder if the difference between the two cases is that in "KANSL1-ARL17A/KANSL1-ARL17B", the alignment corresponding to KANSL1-ARL17A fusion are all discarded, thus no reads support KANSL1-ARL17A (or reads...