Convert untar, pigz/compress, kallisto/index, samtools/index, bowtie2/build to topic versions

[pinin4fjords]

Summary

Converts 5 modules from versions.yml file-based version emission to the modern topic: versions channel emission pattern:

• untar
• pigz/compress
• kallisto/index
• samtools/index
• bowtie2/build

Changes per module

• main.nf: Replace path "versions.yml", emit: versions with structured tuple output using topic: versions
• main.nf: Remove heredoc version blocks from both script: and stub: sections
• meta.yml: Remove old versions: output section, add versions_<tool>: output and topics: section
• tests/main.nf.test: Update assertions to use process.out.findAll { key, val -> key.startsWith('versions') }
• tests/main.nf.test.snap: Regenerated with structured version tuples ["PROCESS_NAME", "tool", "version"]

Version output format

Before:

path "versions.yml", emit: versions

After:

tuple val("${task.process}"), val('toolname'), eval('tool --version ...'), emit: versions_toolname, topic: versions

Bug fix: fastq_align_dedup_bwamem subworkflow

Fixed non-deterministic GPU test failures caused by incorrect use of .first() on subworkflow version channels:

Problem: The workflow used .first() on BAM_SORT_STATS_SAMTOOLS.out.versions and FASTQ_ALIGN_BWA.out.versions, which are subworkflow outputs containing multiple versions mixed from parallel processes (FLAGSTAT, IDXSTATS, etc.). The .first() operator randomly selected whichever version emerged first from the channel, causing:

• Single md5 values to flip between FLAGSTAT and IDXSTATS versions across runs
• GPU tests to fail non-deterministically due to variable process timing

Fix:

• Remove .first() from subworkflow version channels (should only be used on process outputs, not subworkflows)
• Sort versions by md5 hash in tests for deterministic ordering

Conda environment fixes

Pin dependencies in several modules to match their container versions, preventing compatibility issues when running with conda (particularly numpy 2.x and pandas dtype coercion issues):

• gmmdemux: Pin Python 3.12.2, numpy 1.26.4, pandas 2.2.1, scipy 1.12.0, scikit-learn 1.4.1
• pypgx (all 4 modules): Pin Python 3.12.3, numpy 1.26.4, pandas 2.2.2, scipy 1.13.1, scikit-learn 1.5.0
• emu/abundance: Pin Python 3.13.9, numpy 2.3.4, pandas 2.3.3
• orthofinder: Pin Python 3.12.1, numpy 1.26.3, scipy 1.11.4
• checkm2/predict: Pin Python 3.12.9, numpy 1.26.4, pandas 2.2.3, scipy 1.15.2, scikit-learn 1.6.1, tensorflow 2.17.0, keras 3.9.0

Test plan

• All modules pass nf-core modules lint
• All tests pass locally (except kallisto non-stub which requires x86_64 - snapshot manually updated)
• CI tests pass

🤖 Generated with Claude Code

[mashehu]

This PR is getting quite big, maybe split them up by tool?

[pinin4fjords]

This PR is getting quite big, maybe split them up by tool?

It was only 5 modules! The fallout is just extensive. If have to make a PR per module I'm not going to be very motivated to tackle these updates.

[mashehu]

The fallout is just extensive

That's why I would split it up into per module (or per tool) PR, to make it also easier to finish them

[pinin4fjords]

The fallout is just extensive

That's why I would split it up into per module (or per tool) PR, to make it also easier to finish them

I've got 35 of these to do 😢

[mashehu]

you mean <1,500 😜 https://nf-co.re/stats/code/container_conversion/

[pinin4fjords]

This PR is getting quite big, maybe split them up by tool?

It was only 5 modules! The fallout is just extensive. If have to make a PR per module I'm not going to be very motivated to tackle these updates.

you mean <1,500 😜 https://nf-co.re/stats/code/container_conversion/

Ha! Dream on 😛

[pinin4fjords]

@nf-core-bot update gpu snapshot path: subworkflows/nf-core/fastq_align_dedup_bwamem/tests/gpu.nf.test

[pinin4fjords]

@nf-core-bot update gpu snapshot path: subworkflows/nf-core/fastq_align_dedup_bwameth/tests/gpu.nf.test

[SPPearce]

Be nice to shorten this expression, but not going to worry about it here.

[pinin4fjords]

Yeah, I'd rather leave expressions as they were before the topics change if poss.

[pinin4fjords]

There's some issue with the Galah test w/ Docker in CI today that wasn't there on Friday. It relies on a file on zenodo, and maybe the CI runners are disallowed.

But it passes on a clean linux machine I started:

ubuntu@ip-172-31-32-102:~/modules$ nf-test test modules/nf-core/galah/tests/main.nf.test --profile docker --verbose

🚀 nf-test 0.9.3
https://www.nf-test.com
(c) 2021 - 2024 Lukas Forer and Sebastian Schoenherr

Load .nf-test/plugins/nft-anndata/0.1.0/nft-anndata-0.1.0.jar
Load .nf-test/plugins/nft-bam/0.6.0/nft-bam-0.6.0.jar
Load .nf-test/plugins/nft-csv/0.1.0/nft-csv-0.1.0.jar
Load .nf-test/plugins/nft-compress/0.1.0/nft-compress-0.1.0.jar
Load .nf-test/plugins/nft-fastq/0.0.1/nft-fastq-0.0.1.jar
Load .nf-test/plugins/nft-utils/0.0.7/nft-utils-0.0.7.jar
Load .nf-test/plugins/nft-vcf/1.0.7/nft-vcf-1.0.7.jar

Test Process GALAH

  Test [250843e1] 'genomes - no qc_table' 
    > N E X T F L O W  ~  version 25.10.3
    > Launching `/home/ubuntu/modules/.nf-test-250843e1bfa0e25f378392bb851be0a.nf` [sick_neumann] DSL2 - revision: a04438af3e
    > WARN: There's no process matching config selector: CHECKM2_PREDICT
    > WARN: There's no process matching config selector: GAWK
    > [fb/2a5aa6] Submitted process > GALAH (test)
    PASSED (5.96s)
  Test [7c0102a5] 'genomes - checkm2 qc_table' 
    > N E X T F L O W  ~  version 25.10.3
    > Launching `/home/ubuntu/modules/.nf-test-7c0102a5180b51fac4608c4f0036a39d.nf` [dreamy_brown] DSL2 - revision: b7f789018d
    > Staging foreign file: https://zenodo.org/records/14897628/files/checkm2_database.tar.gz
    > [0c/970c16] Submitted process > UNTAR (checkm2_database.tar.gz)
    > [6a/8a1026] Submitted process > CHECKM2_PREDICT (test_checkm2)
    > [0f/785f47] Submitted process > GAWK (test_checkm2)
    > [8f/b2ee16] Submitted process > GALAH (test)
    PASSED (448.54s)
  Test [99a8b8c5] 'genomes - stub' 
    > N E X T F L O W  ~  version 25.10.3
    > Launching `/home/ubuntu/modules/.nf-test-99a8b8c53b6aaed2546b53281583255d.nf` [adoring_lalande] DSL2 - revision: 61228c27d2
    > WARN: There's no process matching config selector: CHECKM2_PREDICT
    > WARN: There's no process matching config selector: GAWK
    > [31/ccd739] Submitted process > GALAH (test)
    PASSED (5.342s)


SUCCESS: Executed 3 tests in 459.855s