Summary

nf-seqlab-progress adds an automatic, hierarchical progress dashboard to nf-seqlab. It combines Nextflow task lifecycle events with structured progress snapshots from native tools to show:

• the current pipeline stage;
• completed source files and total source files;
• active files, phases, and within-file percentages;
• cached, retried, failed, and indeterminate work.

The compact nf-seqlab wordmark and dashboard appear automatically in an interactive terminal. Redirected output, CI, and agent environments receive immutable plain-text status lines instead of cursor control sequences.

Get Started

Pin the plugin in nextflow.config:

plugins {
    id 'nf-seqlab-progress@0.1.0'
}

Import its registration functions in the pipeline entry point:

include {
    registerProgressInputs
    registerProgressStages
} from 'plugin/nf-seqlab-progress'

No wrapper command or separate progress process is required. A normal nextflow run uses the animated dashboard when the terminal supports it.

Examples

Register the complete source-file set before launching tasks, then map process names to user-facing stages:

workflow {
    registerProgressInputs([
        [file_id: 'chr1', path: '/data/chr1.vcf.gz'],
        [file_id: 'chr22', path: '/data/chr22.vcf.gz'],
    ])

    registerProgressStages(
        [
            [id: 'build_svar2', label: 'Build SVAR2', file_ids: ['chr1', 'chr22']],
            [id: 'build_gvl', label: 'Build GVL', file_ids: ['chr22']],
        ],
        [
            [process: 'SEQLAB_BUILD_SVAR2', stage: 'build_svar2', completion_boundary: true],
            [process: 'SEQLAB_NORMALIZE', stage: 'build_gvl', completion_boundary: 'parent'],
            [process: 'SEQLAB_BUILD_GVL', stage: 'build_gvl', completion_boundary: true],
        ],
    )
}

Normal nf-seqlab modules participate automatically when their TaskRun context contains a meta map. File identity resolves from meta.file_id ?: meta.id, and parent identity resolves from meta.parent_file_id ?: meta.parent_id ?: fileId. Optional managed environment inputs remain authoritative when a process provides them directly.

Native snapshot producers export managed values inside their scripts. These shell-local exports are consumed by the producer, while the observer derives the same task ID from the Nextflow work directory:

script:
"""
export NF_SEQLAB_PROGRESS_FILE_ID="${meta.file_id}"
export NF_SEQLAB_PROGRESS_PARENT_FILE_ID="${meta.parent_file_id ?: meta.file_id}"
export NF_SEQLAB_PROGRESS_TASK_ID="\$(basename "\$(dirname "\$PWD")")/\$(basename "\$PWD")"
export NF_SEQLAB_PROGRESS_ATTEMPT="${task.attempt}"
"""

Native tools atomically replace .nf-seqlab-progress.json in the task work directory. A valid snapshot uses the versioned protocol:

{
  "schema": "nf-seqlab.progress/v1",
  "run_id": "focused-curie",
  "stage_id": "build_svar2",
  "process": "SEQLAB_BUILD_SVAR2",
  "file_id": "chr22",
  "parent_file_id": "chr22",
  "task_id": "ed/89cec8...",
  "attempt": 1,
  "state": "running",
  "phase": "read",
  "completed": 4409063557,
  "total": 44090635573,
  "unit": "compressed_bytes",
  "percent": 10.0,
  "message": "Reading variants",
  "updated_at": "2026-07-15T03:34:00Z"
}

License

Apache License 2.0. See COPYING.