Binding for containers

Ask for help
1

I have a very particular use case. I have a very simple script that moves directories from one part of the file system to another (they are mostly sequencing runs) and updates a database, I put that as a nextflow pipeline and my plan was to use that on the seqera platform to move data (it’s a long story about why we need to do that). I can do this using a conda environment. However, if I try to use a container then the filesystem areas that I need are not bound by default. How can I do that? Do you have a suggestion for this use case? My sample sheet would be something like:

source_directory, target_directory, principal_investigator

I specify the above attributes as “vals” as this really involves moving things around the file system and it seems different than what I usually do (start with fastqs for example, and then doing a chain of commands to them, which would end with an output). Do you have any particular suggestions for this?

2

You can often add bind options using the runOptions property of your container scope.

e.g.

docker {
    enabled = true
    runOptions = '-v /path/outside:/path/inside'
}

It seems a bit strange though to be using Nextflow for this kind of processing. With Nextflow, ideally you should work in the work directory, and use the publishing syntax to then put things in their new location. Or is that what you’re doing? Manipulating folder structure in the work directory, and then using the target_directory with publishDir?