Why?

To decrease number of total files in the results/ directory of the job.

How?

With a bash ugly-liner, something like this

for dir in `find . -maxdepth 1 -type d  | grep -v "^\.$" `; do tar -cvzf ${dir}.tar.gz ${dir} && rm -R ${dir}; done

Which reads like this:

Search for folders in the working directory (excluding itself), create a .tar.gz of every folder and remove the initial folder.

Try it with this reprex:

mkdir folder1 folder2
cd folder1 && touch file1 file2 && cd -
cd folder2 && touch file1 file2 && cd -

this should be part of the rmats.nf pipeline to download a requested version of annotation or to have the user supply their own (with long read RNA-sequencing) this is the case. That is, that the user will make a sample specific transcriptome, in the form of a GTF file, this is the outcome of the https://github.com/TheJacksonLaboratory/LyRICS pipeline.
ftp://ftp.ebi.ac.uk/pub/databases/gencode/Gencode_human/release_33/gencode.v33.annotation.gtf.gz

This is a transferred issue initially posted here: TheJacksonLaboratory/sbas#53

This public repository would benefit from a pre-amble, documentation, continuous integration and development tests, etc....

When rMATS (any version) is run, the fromGTF files define the junction definitions for each of the splicing types (A3SS, A5SS, RI, SE, MXE) and junction counting types (JC, JCEC) and counting types (IJC, SJC, INC (calculated), INCLEN, SKIPLEN) the saveSampleCounts.sh has some good documentation which should go into a README.md for matrix formation.

Quoting Anne above.

• Do we have the script she mentions named saveSampleCounts.sh
• Do we have as ouputs the fromGTF.{<splicing-event-type>}.txt files to be able to use?

The latter are input files for the sbas repo scripts.

Referencing an issue posted in TheJacksonLaboratory/sbas#79 (comment) but is relevant to the nf pipeline here.