Does anyone know if Seqera is write in saying this > TSV Alternative: The TSV format is often more reliable and easier to work with for tables in MultiQC. I'm revisting trying to get the full parameter list listed in an (ideally) hidden list at teh bottom of teh table. I wanted to exploit this: https://github.com/nf-core/tools/blob/fee44dae3fb1f86561f8ee11a62eb6207804dc5d/nf_core/pipeline-template/subworkflows/nf-core/utils_nextflow_pipeline/main.nf#L73-L82 which produces

{
    "input": "<https://raw.githubusercontent.com/nf-core/test-datasets/funcscan/samplesheet_reduced.csv>",
    "multiqc_config": null,
    "multiqc_title": null,
    "multiqc_logo": null,
    "max_multiqc_email_size": "25.MB",

To make a MultiQC compatible JSON as well (my idea was to just inject the MultiQC header info), but this appears to be MUCH more involved to modify the default nf-core parameter JSON to get it into the correct 'long' format of one column for parameter name and one for values than I had hoped... For example: If I just add the header section, like this

bash
{
  "id": "basic_pipeline_params",
  "section_name": "Basic Pipeline Parameters",
  "description": "Core configuration parameters",
  "plot_type": "table",
  "pconfig": {
    "id": "basic_params_table",
    "title": "Basic Configuration"
  },
  "data": {
    "input": "<https://raw.githubusercontent.com/nf-core/test-datasets/funcscan/samplesheet_reduced.csv>",
    "outdir": "./results",
    "annotation_tool": "pyrodigal",
    "run_amp_screening": "true",
    "run_arg_screening": "true",
    "run_bgc_screening": "false"
  }
}

I get a long traceback error. To get it to work, I have to add an intermedia 'key' ( guess normally representing samples):

{
  "id": "pipeline_parameters",
  "section_name": "Pipeline Parameters",
  "description": "Configuration parameters used in the pipeline run",
  "plot_type": "table",
  "pconfig": {
    "id": "pipeline_params_table",
    "title": "Pipeline Configuration",
    "col1_header": "Parameter"
  },
  "data": {
    "Pipeline Configuration": {
      "input": "<https://raw.githubusercontent.com/nf-core/test-datasets/funcscan/samplesheet_reduced.csv>",
      "multiqc_config": "null",
      "multiqc_title": "null",
      "multiqc_logo": "null",
      "max_multiqc_email_size": "25.MB",
      "multiqc_methods_description": "null"
    }
  }
}

But this is nasty because it ends up with a single row wide table with the sample name as 'Pipeline configuratio'n and each parameter is a column To make it wide table I have do this large structure change making each parameter it's own key...

{
  "id": "pipeline_parameters",
  "section_name": "Pipeline Parameters",
  "description": "Configuration parameters used in the pipeline run",
  "plot_type": "table",
  "pconfig": {
    "id": "pipeline_params_table",
    "title": "Pipeline Configuration"
  },
  "data": {
    "input": {
      "Value": "<https://raw.githubusercontent.com/nf-core/test-datasets/funcscan/samplesheet_reduced.csv>"
    },
    "multiqc_config": {
      "Value": "null"
    },
    "multiqc_title": {
      "Value": "null"
    },
    "multiqc_logo": {
      "Value": "null"
    },
    "max_multiqc_email_size": {
      "Value": "25.MB"
    },

Seqera AI keeps telling me make it into a TSV table instead... (which indeed works straight out)

# id: 'pipeline_parameters'
# section_name: 'Pipeline Parameters'
# description: 'Configuration parameters used in the pipeline run'
# plot_type: 'table'
# pconfig:
#   id: 'pipeline_params_table'
#   title: 'Pipeline Configuration'
#   col1_header: 'Parameter'
Parameter	Value
input	<https://raw.githubusercontent.com/nf-core/test-datasets/funcscan/samplesheet_reduced.csv>
multiqc_config	null
multiqc_title	null
multiqc_logo	null
max_multiqc_email_size	25.MB
multiqc_methods_description	null

*So: should I just drop the idea of using the JSON file and converting to TSV instead? *

So your question boils down to:

Should I write a thing using loads of code, or a small amount of code?

YEs and no 😅

To make it wide table I have do this large structure change making each parameter it's own key...

This sounds like it should be a single

map

call, no?

I thought it would be but it wasn't when I tried before... but maybe I didn't persist enough 😅 (I was getting annoyed at Seqera AI at the time not being helpful 😆 )

np! 👍🏻

I had a poke around the custom content code. One gotcha with the JSON route: remember to set

col1_header

in

pconfig

, otherwise the first column defaults to "Sample Name":

"pconfig": {
  "col1_header": "Parameter"
}

TSV gets this for free from the header row.

TSV does sound easier at this point