A community effort to collect a curated set of analysis pipelines built using Nextflow.

nf-core

Hello After the release my full tests are failing from the same reason <https://cloud.seqera.io/orgs/nf-core/workspaces/AWSmegatests/watch/48Q9Hyufz3RrAr/logs> . I dont really understand since they were working in my local.

Screenshot 2026-04-23 at 12.01.08.png

File is definitely missing in the work dir:

how are you staging that one in in the pipeline code?

```    MERGE_REPORTS.out.summary
        .map { meta, file -&gt; tuple([vartype: params.variant_type] + [id: meta.benchmark_tool], file) }
        .set { summary }

    // add path to csv file to the datavzrd input
    summary
        .map { meta, summary_file -&gt;
            def updated_meta = meta + [ csv: summary_file.toString() ]
            def template_file = file("${projectDir}/assets/datavzrd/${meta.id}.datavzrd.template.yaml", checkIfExists: true)
            [ updated_meta, template_file ]
        }
        .set { template_ch }

    CREATE_DATAVZRD_INPUT (
        template_ch
    )

    DATAVZRD (
        CREATE_DATAVZRD_INPUT.out.file_out.join(template_ch)
    )```

I still dont understand. DATAVZRD is raising the error, while CREATE_DATAVZRD_INPUT finishes successfully.

I am also trying to resume the task there but Platform raises and error as fasta and fai files are missing. But those are in the igenomes config so user doesnt need to define if --genome is given (very similar to many pipelines)

I have a couple minutes to look at the code. Let me see if I can spot anything. This smells like something where we can make implicit assumptions of where data lives for local compute which breaks on cloud

Can you validate with a .dump() what is in the output of the Create process and the template_ch?

or maybe even better just the complete input to `CREATE_DATAVZRD_INPUT.out.file_out.join(template_ch)`

For the fasta/fai stuff, the reason it complains is, that you said it is required here: <https://github.com/nf-core/variantbenchmarking/blob/8b21c01749c4447b285d242a198127736f3ffe51/nextflow_schema.json#L197>

Platform at this point doesn't know about igenomes because this is all set later in the pipeline code, it just sees the json file. really, I am guessing if this is all set via the igenomes profile, I am guessing either genome or fasta is required but not both ? (This works a little differently when you submit via the API which is what we do for the full size tests from GHA).

```[[vartype:small, id:rtgtools, csv:/omics/groups/OE0608/internal/kubran/variantbenchmarking-2/work/dd/fea9149b02bd560dfc28d37d29d71c/rtgtools.summary.csv], /omics/groups/OE0608/internal/kubran/variantbenchmarking-2/work/bd/aec7eca47f8cbed73cbe30de448220/rtgtools.datavzrd.yaml, /omics/groups/OE0608/internal/kubran/variantbenchmarking-2/assets/datavzrd/rtgtools.datavzrd.template.yaml]```

I guess trouble is to use the file through meta

yes I think it is staged in as a string and not as proper file. This is fine on local stuff but not on cloud where we need nextflow to do the file handleing for various providers

I see, I will fix it for a patch release :face_exhaling: