Open angarb opened 3 years ago
Suggestion: add documentation in the repo docs on how to run rMATS stats standalone from the pipeline, for example by using the pipeline's container, or an installed tool. Provide examples how to do that on the pipeline's output + additional external samples.
@mkostich has worked on this. We must integrate it into the pipeline:
Notes: "I forked your repo and edited the fork. That way I didn't have to worry about permissions issues. We can merge the fork back into your repo when you are happy with it.
Fork is here: https://github.com/mkostich/splicing-pipelines-nf Branch with changes is 'mitch1' here: https://github.com/mkostich/splicing-pipelines-nf/tree/mitch1 I didn't edit the docs or the wiring diagram. Want to make sure it works for you first. Test data seemed to work on Sumner, but we are still fiddling with GCP. Also, needs testing with single-ended short-read data. Two-stage rMATS is only implemented for when you supply the rmats_pairs (.txt) file. We got pulled in other directions, but hope to get back to the GCP aspect next week.
install and get right branch: git clone https://github.com/mkostich/splicing-pipelines-nf.git git checkout mitch1
edit line 11 in main.pbs to match your install location: project_dir=/path/to/splicing-pipelines-nf
copy 'my.cfg' (equivalent of NF_splicing_pipeline.config, just more convenient name) to working directory (where you launch nextflow), edit it there (make sure to check assembly_name, readlength and minlen; ensure minlen < readlength to allow for adapter removal), and then launch the pipeline:
sbatch /path/to/splicing-pipelines-nf/main.pbs my.cfg
I created a pull request for the changes hereL #327
Problem
A new parameter --task stat has been added to rMATS. This will allow us to do statistics separately from calculating the PSI values.
Solution
Sangram checked the option
--task
for upgrading the rmats options, docs - https://github.com/Xinglab/rmats-turbo/blob/v4.1.1/README.md#running-prep-and-post-separatelySeems a bit complex, this need a bit of though on how separate both
prep
andstats
into two different processesImplementation
We need to brainstorm. The best option might be to design a
prep
parameter and run thestats
parameter with the second half of the pipeline (or rMATS alone).