Hello everyone. I am pretty new for nextflow and now I have a question. If I initially analyze my data using the RNAseq pipeline and later want to examine RNA splicing, what should I do to use some output of RNAseq piplines? I noticed that the initial steps involving Star and Salmon are quite similar. Thank you.

What I did was invoke the pipeline from the same directory, using the -resume flag. Nextflow will see the previous work directory, and any identical outputs will be considered cached. I think (but could be wrong) that RNA seq and RNA splice using slightly different paraments for alignments, so I think it still runs STAR and salmon, but it will definitely skip the trimming and fastqc. Just make sure your sample sheet and contrasts input are formatted to the RNA splice pipeline properly

@Brian_C Thank you. I think I should try the -resume flag to test. Trimming and fastqc consume limited computational resources, but STAR consumes a lot. But I think theoretically the two piplines should be able to share the output of one STAR.

It is possible to optimize the STAR parameters for what you want to do, e.g. splicing analysis (e.g. rMATS recommends end-to-end alignment, i.e. no clipping) or gene quantification. I have not compared the pipelines and don't know if they do that, but it could be a reason for differences. You still could reuse BAM files, but that is currently not supported in either pipeline, although there should be a GitHub issue somewhere that requests starting rnaseq from BAMs. The problem is that the BAM files have to have certain attributes to work with the downstream tools, so ideally they are created by the pipeline to make sure of that, so while it should be possible to implement, it is not as trivial as it may seem. (If a developer allows something that users can misuse, they will pester him/her with all the error ensuing from that misuse, which takes time away from development work, so it is understandable they want some safeguards against that.)

@Thomas Danhorn Thanks. I think the safest option at this point is still to perform a separate pipeline content.

According to the docs, rnasplice supports multiple inputs, including genome BAMs, transcriptome BAMs, and salmon outputs. https://nf-co.re/rnasplice/dev/docs/usage/#samplesheet-input

Cool! I somehow overlooked that.