Summary

nf-snapshot checks the files a Nextflow run publishes against a snapshot file committed to the pipeline repository, the way nf-test snapshots work, but for one full-size run. If the outputs differ, the run exits with status 1.

Get Started

Declare the plugin at the top level of nextflow.config, then enable it in a profile for each test that has a snapshot. Nextflow 26.10.0 did not load the plugin when it was declared inside a profile. The plugin does nothing unless snapshot.enabled is set.

plugins { id 'nf-snapshot@0.2.0' }

profiles {
    test_full {
        snapshot {
            enabled  = true
            file     = 'tests/snapshots/test_full.json'
            ignore   = ['pipeline_info/**']
            nameOnly = ['**/*.html', '**/*.log']
        }
    }
}

Create the snapshot, then commit it:

NXF_SNAPSHOT_UPDATE=true nextflow run . -profile test_full

Later runs verify against it:

nextflow run . -profile test_full

To accept existing outputs without re-running, run from the project directory:

nextflow plugin nf-snapshot:update --profile test_full --outputDir results

Examples

If the outputs differ, the run prints a diff and fails:

nf-snapshot: outputs differ from /path/to/tests/snapshots/test_full.json
  missing:    salmon/sample2/quant.sf
  changed:    multiqc/multiqc_data/multiqc_general_stats.txt
  unexpected: salmon/sample3/quant.sf

Snapshot format example:

{
  "files": {
    "multiqc/multiqc_report.html": null,
    "salmon/sample1/quant.sf": "9e107d9d372bb6826bd81d3542a419d6",
    "star/sample1.bam": "size:104857600"
  }
}

Each entry holds the MD5 of the file content, size:<bytes> for size-only files, or null for name-only files.

License

Licensed under the Apache License, Version 2.0.