Replace alevinqc with qcatch for simpleaf QC by an-altosian · Pull Request #520 · nf-core/scrnaseq

an-altosian · 2026-02-14T00:34:34Z

Update simpleaf modules to 0.19.5
Add qcatch module from local modules repo
Add qcatch chemistry mappings to protocols.json
Update simpleaf subworkflow to use QCATCH
Add skip_qcatch parameter for optional QC
Remove alevinqc module and bin/alevin_qc.r
Update test config and snapshots

PR checklist

- Update simpleaf modules to 0.19.5 - Add qcatch module from local modules repo - Add qcatch chemistry mappings to protocols.json - Update simpleaf subworkflow to use QCATCH - Add skip_qcatch parameter for optional QC - Remove alevinqc module and bin/alevin_qc.r - Update test config and snapshots Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

an-altosian · 2026-02-14T00:34:58Z

I am almost done with this PR, just need to grab a dataset to test. This can be a breaking change because I replaced alevinqc with qcatch

nf-core-bot · 2026-02-14T00:35:07Z

Warning

Newer version of the nf-core template is available.

Your pipeline is using an old version of the nf-core template: 3.5.1.
Please update your pipeline to the latest version.

For more documentation on how to update your pipeline, please see the nf-core documentation and Synchronisation documentation.

The qcatch module is not yet in the official nf-core modules repo, so having it under modules/nf-core/ with a placeholder git_sha caused the nf-core linter to crash with IndexError. Moving it to modules/local/ avoids the version check. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

fmalmeida · 2026-02-14T08:13:39Z

Hi there,

Thanks for the PR.

Do we have an issue that was tracking the request?

And the testing dataset can be the same that was being used for alevin, no?

fmalmeida

Adding some ideas :)

modules/local/qcatch.nf

fmalmeida · 2026-02-14T08:17:51Z

modules/local/qcatch.nf

+    prefix = task.ext.prefix ?: "${meta.id}"
+
+    """
+    export MPLCONFIGDIR=./tmp


These env variables maybe can be set directly in the nextflow.config as env there so all python modules already use it by default.

Instead of adding in each, no?

modules/local/qcatch.nf

an-altosian · 2026-02-14T14:33:33Z

Thank you @fmalmeida! I will address your comments. BTW, I also mde a PR for nf-core qcatch module nf-core/modules#10032 . I will update there as well.

For the test dataset, as qcatch does cell calling internally (we reverse engineered cellranger's cell calling algorithm) and exports a filtered count matrix, it only works on real datasets. I tried to use the test dataset we have, but it gave me error saying insufficient data for cell calling. Therefore, I end up with doing the same thing we did for cellblender.

In the nf-core/module, I have tests using real data we used in the qcatch github repo. As we will pull the module directly from nf-core/modules, I think we are good even if we don't have tests for qcatch here.

… config - Replace modules/local/qcatch.nf with modules/nf-core/qcatch - Move env vars (MPLCONFIGDIR, NUMBA, etc.) to nextflow.config - Update simpleaf subworkflow import path Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

an-altosian · 2026-02-14T15:27:32Z

All comments resolved. I will find a dataset to test.

The global env vars (TMPDIR, MPLCONFIGDIR, etc.) added for qcatch were breaking other processes: piscem crashed with SIGABRT due to TMPDIR=./tmp, and multiqc plots failed to generate due to MPLCONFIGDIR=./tmp. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

an-altosian · 2026-02-14T16:39:10Z

The failing test seems to be related to STAR.

an-altosian · 2026-02-14T22:31:55Z

Tested. Ready to merge

fmalmeida · 2026-02-16T11:40:25Z

Thank you @fmalmeida! I will address your comments. BTW, I also mde a PR for nf-core qcatch module nf-core/modules#10032 . I will update there as well.

For the test dataset, as qcatch does cell calling internally (we reverse engineered cellranger's cell calling algorithm) and exports a filtered count matrix, it only works on real datasets. I tried to use the test dataset we have, but it gave me error saying insufficient data for cell calling. Therefore, I end up with doing the same thing we did for cellblender.

In the nf-core/module, I have tests using real data we used in the qcatch github repo. As we will pull the module directly from nf-core/modules, I think we are good even if we don't have tests for qcatch here.

Maybe we should focus on merging the nf-core/modules first, because there we will have automated testing for the module. So we then bring it here to this PR ... this would help with the testings, since the pipeline small test cannot be used.

Any other comments @grst and @nictru ?

- Update qcatch module to 0.2.10 (pip-based install with scikit-image) - Add remove_doublets pipeline parameter for Scrublet doublet detection - Replace beforeScript with global env vars in nextflow.config - Register qcatch as nf-core module in modules.json - Update test snapshots for deterministic 0.2.10 outputs Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

DongzeHE · 2026-02-21T21:23:32Z

I want to discuss one thing: qcatch provides emptydrop and doublet removal. Therefore, cellbench is not needed when running simpleaf mode. How to we want to expose it? BTW, the qcatch report is pretty cool. Check this demo out https://combine-lab.github.io/QCatch/demo/demo.html

…itespace Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

fmalmeida · 2026-02-25T15:25:25Z

When you say provide, is that an option or something that is always done?

If the latter then yeah, we would end up having to remove cellbender from alevin runs.

If it is optional, then we can make the PR simpler and split the work where here we just bring qcatch and keep cellbender as the key module for this, opening a new issue to have this bigger discussion around how to best handle it.

@grst any takes?

an-altosian · 2026-02-25T15:28:29Z

The empty droplet removal is mandatory (thanks to our reviewer No.2 ;) ) and the doublet correction is an optional. I added a parameter in my PR called remove_doublets to turn this on.

grst · 2026-02-25T15:33:23Z

cellbender is anyway optional and disabled by default.
If possible, I would still like to give users the option to use cellbender with simpleaf, in case they believe it's superior for whatever reason.

btw, is QCatch specific to simpleaf of would that also work with the other aligners in the pipeline?

an-altosian · 2026-02-25T15:36:11Z

It is designed for only simpleaf and alevin-fry currently. Extending the support to other tools are our future plans.

The updated qcatch container (with scikit-image) produces different h5ad/seurat checksums for Sample_X due to changed package metadata. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

DongzeHE requested review from fmalmeida, grst and nictru and removed request for fmalmeida and grst February 14, 2026 01:03

fmalmeida reviewed Feb 14, 2026

View reviewed changes

an-altosian and others added 2 commits February 14, 2026 15:42

Add only required env vars for qcatch Singularity compatibility

860bd29

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

an-altosian marked this pull request as ready for review February 14, 2026 22:31

Sync qcatch module with merged nf-core/modules#10125, fix trailing wh…

3e32fb6

…itespace Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Merge branch 'dev' into dev

aa04064

Update simpleaf pipeline test snapshot for new qcatch container

2460e51

The updated qcatch container (with scikit-image) produces different h5ad/seurat checksums for Sample_X due to changed package metadata. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

Conversation

an-altosian commented Feb 14, 2026

PR checklist

Uh oh!

an-altosian commented Feb 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

nf-core-bot commented Feb 14, 2026

Uh oh!

fmalmeida commented Feb 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

fmalmeida left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

fmalmeida Feb 14, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

an-altosian commented Feb 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

an-altosian commented Feb 14, 2026

Uh oh!

an-altosian commented Feb 14, 2026

Uh oh!

an-altosian commented Feb 14, 2026

Uh oh!

fmalmeida commented Feb 16, 2026

Uh oh!

DongzeHE commented Feb 21, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

fmalmeida commented Feb 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

an-altosian commented Feb 25, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

grst commented Feb 25, 2026

Uh oh!

an-altosian commented Feb 25, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

an-altosian commented Feb 14, 2026 •

edited

Loading

fmalmeida commented Feb 14, 2026 •

edited

Loading

an-altosian commented Feb 14, 2026 •

edited

Loading

DongzeHE commented Feb 21, 2026 •

edited

Loading

fmalmeida commented Feb 25, 2026 •

edited

Loading

an-altosian commented Feb 25, 2026 •

edited

Loading