extra_salmon_quant_args quote issue

[kdivilov]

Description of the bug

Setting --extra_salmon_quant_args "--noLengthCorrection" leads to the following error message:
The following invalid input values have been detected: * --extra_salmon_quant_args: expected type: String, found: Boolean (true)

I believe the reason for this error is that the quotes surrounding --noLengthCorrection are removed somewhere along the way. This is based on there being no quotes around it in the DEBUG nextflow.cli.Launcher line in the log file.

Command used and terminal output

No response

Relevant files

No response

System information

Nextflow version 23.10.0
nf-core/rnaseq 3.14.0
singularity

[pmoris]

I just encountered this issue myself while working on a custom pipeline that uses the same extra_args option as the rnaseq pipeline. From what I can tell, the problem only appears when providing a single extra argument, but it works fine for multiple ones.

For example:

--fastp_extra_args '--trim_poly_x' # interpreted as --trim_poly_x: true at the nextflow level
--fastp_extra_args '--trim_poly_x -y' # both args are interpreted by fastp only

The modules.config entry for the FASTP module contains this line:

params.fastp_extra_args ?: ''

And the parameter was also made available in the nextflow.config file.

Single and double quotes appear to behave similarly.

All of the above applies to the CLI, I did not test the behaviour when using a params.yml file.

[pinin4fjords]

This is something to do with nf-validation I think. I'll dig into that, and maybe #1290 will help. I was wrong- this is the CLI, see nf-core/fetchngs#308 (comment).

For now, I've noticed that prepending the string with a space also forces the string interpretation:

 --extra_salmon_quant_args " --noLengthCorrection"

[pinin4fjords]

https://nfcore.slack.com/archives/C056RQB10LU/p1717083898763659 to follow possible resolution via nf-validation

[ewels]

Also I think using an = should be another short term fix:

 --extra_salmon_quant_args="--noLengthCorrection"

[pinin4fjords]

Thanks @ewels !

Closing this now, since it's not an rnaseq issue and possibilities for full resolution lie elsewhere.