Thanks a lot, that was exactly what I was looking for !
Hi,
I've written a pipeline to perform quantification from RNAseq data with salmon.
I'm trying to find a way to evaluate the quality of my results.
I was thinking to run the pipeline on available public dataset and compare my output with another analysis.
Would anyone have heard of an RNAseq dataset for which the fastq are available, and that have been quantified with salmon?
Thanks
1 answer
I think it is important that we validate results one way or another, especially initially when you are learning about the process
What is really important to attempt to do so with different methods, much can be learned about the limitations that way.
I would recommend running a genome alignment-based workflow: hisat2 for example (or star) and see what you get that way.
Then there is also the Recount project an online resource that provides results for thousands of studies for human and mouse respectively.
Pick one study there, or run their pipeline on your results
Log in to answer this question.
Do you really think that is necessary? After all the quantification for the end user comes down to building the index and running salmon itself. If your pipeline is written in a way that the actual command lines that are executed are returned then you can easily see if that is in line with what the manual recommends -- so checking if your pipeline puts that all together correctly. Tbh that would be all I'd do. Check the salmon logs of the run, and if that is ok then go ahead. I always make a minimal testing dataset to see if the pipeline runs well, and run it after code changes to see I did not break anything. GitHub actions is great fot that.