Hi, I have 2 mutually exclusive options in my next...
# random
s
Hi, I have 2 mutually exclusive options in my nextflow.config
image_based
set to true &
coordinate_based
set to false by default. The pipeline should run the image-based analysis workflow by default, but if the user provides the
--coordinate_based
, image_based should be set to false and the pipeline should run the coordinate-based workflow. In this case I am trying to set
params.image_based=false
but this does not work and both the options are now set to true in turn breaking my pipeline run. I want to understand the scope of the params being read from the nextflow.config file. At what stage can I update a value of a param? Is it before the pipeline initialization subworkflow? In that case where should the below code snippet reside?
Copy code
def validateInputParameters() {

    // check if both `--image_based` & `--coordinate_based` are true
    if ( params.image_based && params.coordinate_based ) {
        error "ERROR: Use either `--image_based` or `--coordinate_based` not both."
    } // this works

    // set image_based to false if coordinate based is provided
    if (params.coordinate_based ) {
        params.image_based = false
    } // params.image based is not updated to false

    // check if --relabel_genes is true but --gene_panel is not provided
    if ( params.relabel_genes && !params.gene_panel ) {
        log.warn "Relabel genes is enabled, but gene panel is not provided with the `--gene_panel`. Using `gene_panel.json` in the xenium bundle"
    }

}
a
params
are scoped such that workflows (or modules) have access to the params in the state they were at the point in the code that the workflow/module was
included
Consider the following:
Copy code
params.test = 'a'
include { SUBWF1 } from '../workflows/subworkflow1/main.nf'

params.test = 'b'
include { SUBWF2 } from '../workflows/subworkflow2/main.nf'
In this scenario if code in
SUBWF1
attempts to access
params.test
it will have the value
a
. If code in
SUBWF2
tries to use
params.test
it will instead see
b
. This is one of the many reasons it is not recommended to access
params
directly outside of the entry workflow. It is also why it’s not recommended to try and modify
params
values after their initialisation - it rarely works. In your situation you should most likely do one of a few things: • make the parameters truly mutually exclusive and don’t attempt to allow the user to have one as
default=true
(i.e. user must supply one of the two flags but cannot supply both. They both default to
false
) • make the parameters mutually exclusive within your code logic by always testing them as a pair:
if (params.coordinate_based && !params.image_based)
• use the parameters to set values within the pipeline which can then be passed between workflows/modules:
Copy code
image_based = params.image_based
coordinate_based = params.coordinate_based && !params.image_based

MY_SUBWORKFLOW(some_channel, image_based)
party this 1
t
Arthur gave you an excellent explanation of why your code doesn't work, but I don't see why you would have two parameters to begin with -- this seems like a simple binary decision (you run the pipeline either image-based or coordinate-based, never both, nor a mixture), so a single logical parameter (e.g.
--coordinate_based
) should hold that single bit of information. Whenever it is false, the workflow is image-based, and when it is true, it is coordinate-based. If you foresee adding more modes in the future, you could have a parameter for the mode, either
--mode image
or
--mode coordinate
, or maybe in the future you can add
--mode xyz
. If you do that, check the parameter at the beginning and make sure it only contains valid settings.
party this 1
👍 1
s
Hi Thomas, I have implemented exactly what you have mentioned🙌
👍 1
🤞 1
the reason I went with the earlier implementation is that we would add a
segmentation-free analysis
going forward (in later releases),
image-based
and
coordinate-based
are segmentation approaches