alternative splicing: very different outputs from rMATS vs altanalyze

Question

[smurph50]

We are comparing two RNA-seq datasets (read length = 150nt, depth = 50M reads, 2 conditions, n = 4 per condition) for alternative splicing. I used rMATS and altanalyze to find alternative splicing events. However, there is very little overlap between the results.

This paper says that there wasn't much overlap between rMATS and SUPPA in their study. https://www.nature.com/articles/s41467-020-15815-7#Sec15

We plan on experimentally validating some of the top hits using qPCR. Which results should we use? Are they both correct? Is this due to my error in running rMATS?

[i.sudbery]

Not directly relevant, but this paper: https://bmcbioinformatics.biomedcentral.com/articles/10.1186/s12859-021-04263-9

Found that SUPPA and rMATS are very similar expect when it came to intron retension. As the paper you cited is mostly about intron retension, this could be an explaination of the differences seen.

The altanalyze wiki also notes that the biggest differences come from intron retension events: https://github.com/nsalomonis/altanalyze/wiki/Algorithms#multipath-psi-splicing-algorithm