Add module: dorado/basecaller by sahuno · Pull Request #11122 · nf-core/modules

sahuno · 2026-04-06T02:09:36Z

Summary

Adds DORADO_BASECALLER process for Oxford Nanopore basecalling with dorado
Supports automatic model selection, modified base calling (5mCG_5hmCG, 5mC, m6A), and optional integrated alignment
GPU-accelerated; follows the same container pattern as nf-core/parabricks modules

Key design decisions

Container — no bioconda/BioContainers (ONTPL licence)

dorado is not on bioconda due to its Oxford Nanopore commercial licence. The module uses the official nanoporetech/dorado Docker Hub image directly (SHA-pinned to v1.4.0), the same pattern used by parabricks modules (nvcr.io/nvidia/...). conda null is set with a comment. ONT does not publish semver Docker tags yet — tracked in nanoporetech/dorado#1584; the SHA will be replaced with a version tag when available.

GPU label — process_gpu

The module requires a CUDA GPU and uses label 'process_gpu'. This label is defined in nf-core/configs but is flagged as non-standard by lint. Happy to discuss the right approach on the PR.

CI tests — stub only (GPU not available on GitHub Actions)

2 stub tests run in CI: unaligned BAM and aligned BAM (reference input path)
2 real GPU tests tagged gpu are excluded from CI but pass locally on a SLURM GPU node
Precedent: parabricks modules also skip GPU CI tests

Test data

tests/data/test.pod5 — minimal synthetic pod5 for stub tests
tests/data/HG002_PAW70337_giab_10reads.pod5 — 10-read GIAB HG002 public ONT data (pending Add pod5 test data: HG002 ONT 10-read subset (GIAB PAW70337, 5kHz R10.4.1) test-datasets#1968 for upstream hosting; included locally for GPU test validation)

Lint results

48 tests passed, 0 failures, 4 warnings (all explained above: quay.io 403, container version mismatch, process_gpu label)

PR checklist

🤖 Generated with Claude Code

Oxford Nanopore dorado basecaller module with automatic model selection, modified base calling (5mCG_5hmCG via combined model syntax), and optional alignment to a reference genome. Uses local SIF for Track 1 (lab use); Track 2 upstream blocked pending dorado bioconda package. Tests: 2 stub (CPU) + 2 real GPU tests (NVIDIA L40S, componc_gpu_batch). Uses --profile singularity,gpu to expose GPU via --nv flag. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Adds a stub test using the public HG002 GIAB pod5 from nf-core/test-datasets (PR nf-core/test-datasets#1968). Test references the file via modules_testdata_base_path so it resolves once that PR merges. Also commits the 2.2 MB 10-read pod5 subset locally for development. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Thin wrapper over nanoporetech/dorado:sha... adding version labels. Used by `wave freeze --dockerfile Dockerfile` to obtain a stable community.wave.seqera.io URI for nf-core upstream submission. See sandbox/TRACK2_wave_freeze.md for the full runbook. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

nf-core/parabricks modules use nvcr.io/nvidia/... directly without bioconda. Following same pattern: use nanoporetech/dorado:sha... from Docker Hub. Removes the singularity/docker ternary — single container field only. Local SIF override kept in tests/nextflow.config for MSKCC HPC testing. Tracking semantic version tags: nanoporetech/dorado#1584. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

- Container: sahuno/dorado:1.4.0 (wraps nanoporetech/dorado v1.4.0 + samtools) - meta.yml: document that dorado outputs SO:unknown; advise SAMTOOLS_SORT + SAMTOOLS_INDEX downstream (confirmed with GIAB HG002 10-read test on A100) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

- Correct model input YAML structure (remove extra nesting level) - Add topics: section for version broadcast - Fix licence list format - Add edam ontology for summary TSV output Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Switch from sahuno/dorado:1.4.0 wrapper to the upstream nanoporetech/dorado:shac8f356489fa8b44b31beba841b84d2879de2088e (v1.4.0). Same approach as parabricks modules using nvcr.io/nvidia directly. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

- tests/nextflow.config: remove hardcoded local SIF container override, local models_dir, and /data1/greenbab reference paths - tests/main.nf.test: stub test 2 uses [[], [], []] for reference (stub block doesn't use reference; real aligned test is GPU-tagged); GPU test with reference uses params.modules_testdata_base_path - Remove GIAB Track 2 stub test (pending nf-core/test-datasets#1968) - Update snapshots Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

ext.device is not an allowed ext key in nf-core modules (only ext.args, ext.prefix, ext.when are permitted). Hardcode --device cuda:all directly in the script block. Users needing a different device string can pass it via ext.args (e.g. --device cpu for testing). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

dialvarezs

Hi, thanks for your contribution.

In general this looks fine to me.
An improvement would be requesting to upload the container to the nfcore org.

I have some comments.
And also, you have to remove the test data.

dialvarezs · 2026-04-06T11:17:23Z

modules/nf-core/dorado/basecaller/main.nf

+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        dorado: 1.4.0
+    END_VERSIONS


Suggested change

cat <<-END_VERSIONS > versions.yml

"${task.process}":

dorado: 1.4.0

END_VERSIONS

dialvarezs · 2026-04-06T11:23:26Z

modules/nf-core/dorado/basecaller/main.nf

+    output:
+    tuple val(meta), path("*.bam")      , emit: bam
+    tuple val(meta), path("*_summary.tsv"), emit: summary , optional: true
+    tuple val(meta), path("*.log")      , emit: log       , optional: true


dorado basecaller doesn't have any option to write the log in a file, so it's not possible to get this output.

dialvarezs · 2026-04-06T11:24:18Z

modules/nf-core/dorado/basecaller/main.nf

+    // Docker Hub image directly — same pattern as nf-core/parabricks modules
+    // (nvcr.io/nvidia/...). SHA tag pins to v1.4.0; a semver tag is tracked in


Suggested change

// Docker Hub image directly — same pattern as nf-core/parabricks modules

// (nvcr.io/nvidia/...). SHA tag pins to v1.4.0; a semver tag is tracked in

// Docker Hub image directly. SHA tag pins to v1.4.0; a semver tag is tracked in

dialvarezs · 2026-04-06T11:29:37Z

modules/nf-core/dorado/basecaller/tests/nextflow.config

This config is not used. You're not loading it in main.nf.test.
So, if it's not necessary, then I would remove it.

dialvarezs · 2026-04-06T11:29:48Z

modules/nf-core/dorado/basecaller/main.nf

+    // (nvcr.io/nvidia/...). SHA tag pins to v1.4.0; a semver tag is tracked in
+    // nanoporetech/dorado#1584.
+    conda null
+    container "nanoporetech/dorado:shac8f356489fa8b44b31beba841b84d2879de2088e"


Suggested change

container "nanoporetech/dorado:shac8f356489fa8b44b31beba841b84d2879de2088e"

container "docker.io/nanoporetech/dorado:shac8f356489fa8b44b31beba841b84d2879de2088e"

dialvarezs · 2026-04-06T11:32:15Z

.gitignore

Don't modify this. It shouldn't be necessary.

sahuno · 2026-04-06T21:39:40Z

Thanks for the review @dialvarezs! All points addressed in 8779494:

Container: switched from sahuno/dorado:1.4.0 to docker.io/nanoporetech/dorado:shac8f356489fa8b44b31beba841b84d2879de2088e (official upstream image with docker.io/ prefix)
Comment wording: updated as suggested
*.log output: removed from main.nf, meta.yml, and stub — confirmed dorado basecaller has no --log-file flag
Test data: removed both pod5 files from tests/data/ — the GIAB HG002 pod5 is now hosted in nf-core/test-datasets (merged in Added fq/lint module #1968) and referenced via params.modules_testdata_base_path
tests/nextflow.config: removed — it was not loaded by main.nf.test and contained local paths
.gitignore: no changes made to this file

Regarding the container upload to the nfcore org — happy to pursue that. Should I open a request with ONT to mirror nanoporetech/dorado under docker.io/nfcore/dorado, or is there an existing process for this in nf-core?

Lint: 51 passed, 0 failures (4 expected warnings for Docker Hub / SHA tag / process_gpu label).
Tests: 4/4 stub tests pass.

dialvarezs · 2026-04-07T11:14:22Z

It seems that you forgot to push the changes.

sahuno and others added 8 commits April 5, 2026 21:16

fix: dorado/basecaller lint --fix corrections

b3066a0

- Correct model input YAML structure (remove extra nesting level) - Add topics: section for version broadcast - Fix licence list format - Add edam ontology for summary TSV output Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

sahuno requested review from a team, edmundmiller and maxulysse as code owners April 6, 2026 02:09

dialvarezs requested changes Apr 6, 2026

View reviewed changes

sahuno requested a review from dialvarezs April 6, 2026 21:39

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add module: dorado/basecaller#11122

Add module: dorado/basecaller#11122
sahuno wants to merge 9 commits intonf-core:masterfrom
sahuno:add-dorado-basecaller

sahuno commented Apr 6, 2026

Uh oh!

dialvarezs left a comment •

edited

Loading

Uh oh!

dialvarezs Apr 6, 2026

Uh oh!

dialvarezs Apr 6, 2026

Uh oh!

dialvarezs Apr 6, 2026

Uh oh!

dialvarezs Apr 6, 2026

Uh oh!

dialvarezs Apr 6, 2026

Uh oh!

dialvarezs Apr 6, 2026

Uh oh!

sahuno commented Apr 6, 2026

Uh oh!

dialvarezs commented Apr 7, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

		// Docker Hub image directly — same pattern as nf-core/parabricks modules
		// (nvcr.io/nvidia/...). SHA tag pins to v1.4.0; a semver tag is tracked in

	// Docker Hub image directly — same pattern as nf-core/parabricks modules
	// (nvcr.io/nvidia/...). SHA tag pins to v1.4.0; a semver tag is tracked in
	// Docker Hub image directly. SHA tag pins to v1.4.0; a semver tag is tracked in

	container "nanoporetech/dorado:shac8f356489fa8b44b31beba841b84d2879de2088e"
	container "docker.io/nanoporetech/dorado:shac8f356489fa8b44b31beba841b84d2879de2088e"

Conversation

sahuno commented Apr 6, 2026

Summary

Key design decisions

Test data

Lint results

PR checklist

Uh oh!

dialvarezs left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

dialvarezs Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

dialvarezs Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

dialvarezs Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

dialvarezs Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

dialvarezs Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

dialvarezs Apr 6, 2026

Choose a reason for hiding this comment

Uh oh!

sahuno commented Apr 6, 2026

Uh oh!

dialvarezs commented Apr 7, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

dialvarezs left a comment •

edited

Loading