mirtrace_species variable not defined

[kopichris]

Description of the bug

According to the docs, the --mirtrace_species variable is set when the --genome variable is defined; however, setting the --genome parameter results in the following error:

Reference species for miRTrace is not defined via the `--mirtrace_species` parameter.

The error occurs because the --mirtrace_species variable is not globally defined, and is accessed by a function that does not have access to it in the pipeline_initialisation workflow:

def validateInputParameters() {
    genomeExistsError()

    if (!params.mirgenedb && !params.mirtrace_species) {
            error("Reference species for miRTrace is not defined via the `--mirtrace_species` parameter.")
    }
...

I am working on making a pull request to the dev branch if there are no issues!

Command used and terminal output

$ nextflow run main.nf -profile docker --input samplesheet.csv --outdir debug_test --genome "GRCh38"

 N E X T F L O W   ~  version 24.04.2

Launching `main.nf` [adoring_bell] DSL2 - revision: 96a86dc2ea

[-        ] process > PREPARE_GENOME:CLEAN_FASTA  -
[-        ] process > PREPARE_GENOME:INDEX_GENOME -
Reference species for miRTrace is not defined via the `--mirtrace_species` parameter.

Relevant files

No response

System information

Nextflow version: 24.04.2
Hardware: HPC
Executor: local
Container engine: Docker
OS: Linux
Version of nf-core/smrnaseq: 2.4.0

[nschcolnicov]

Hi @kopichris Thanks for opening this issue, I see that you are right. The validateInput() function is running before these parameters have a chance to be set with params.mirtrace_species = getGenomeAttribute('mirtrace_species').
Also tagging @atrigila who suggested that validateInput() was the one causing this issue.
Please tag me or her on the PR once its ready

[kopichris]

Hi @nschcolnicov, that's what I suspected at first too, but I think it's just a parameter visibility issue.

The params.mirtrace_species is correctly defined and set prior to validateInputParameters() being called - this can be seen on line 31 in main.nf.

The parameter is then passed as input to the prepare_genome workflow, where the following check happens:

if (!params.mirgenedb && !val_mirtrace_species) {
    exit 1, "Reference species for miRTrace is not defined via the --mirtrace_species parameter."
}

A similar check is subsequently performed within the pipeline_initialisation workflow, where the validateInputParameters() function is implemented:

if (!params.mirgenedb && !params.mirtrace_species) {
    error("Reference species for miRTrace is not defined via the `--mirtrace_species` parameter.")
}

The error occurs because params.mirtrace_species was not passed as input to the pipeline_initialization workflow, so params.mirtrace_species is null instead of hsa, for example.

I chose to remove this logic from the pipeline_initialization workflow because it was already being correctly handled in prepare_genome (69ef3aa).

Thoughts @nschcolnicov @atrigila?

[nschcolnicov]

@kopichris I think that change makes sense, not sure why the same check is run twice, probably some leftover code that we missed when refactoring the pipeline. Please go ahead and create the PR. Thanks for looking into this! 😄

[atrigila]

Closed via #493 . Thank you for reporting this issue and the proposed solution :)