New module: modkit/extractcalls

[sahuno]

PR checklist

• This comment contains a description of changes (with reason).
• Stub + real tests added.
• New tool follows module conventions.
• Versions broadcast via `topic: versions`.
• Naming/parameter/I-O conventions followed.
• Resource label set (`process_high`).
• BioConda + BioContainers used.
• `nf-core modules lint modkit/extract/calls` — 49/0/0.
• `nf-test test --profile conda` — 2/2 passed.

Summary

Adds a new nf-core module wrapping `modkit extract calls`, which produces a per-read per-position table of base-modification calls (pass / fail / filtered, with the called base) using the same thresholding algorithm as `modkit pileup`.

Complementary to `modkit/extract/full`: `extract calls` emits the thresholded categorical decision per site per read, while `extract full` emits the underlying probabilities.

The module auto-detects `--bgzf` in `ext.args` and adjusts the output filename suffix accordingly.

Why

`modkit extract calls` is the go-to tool for per-read allele-specific methylation, methylation-aware phasing validation, and read-level QC where you want the same thresholded labels as the pileup output but at read level rather than site level.

Test data

Uses the existing `test.sorted.phased.bam` from nf-core/test-datasets (modules branch). No new test data required.

🤖 Generated with Claude Code

[SPPearce]

This is named incorrectly, it is too nested in the subfolder.
So modkit/foo, but not modkit/foo/bar

[SPPearce]

So modkit/extractcalls here

[github-actions]

This PR has been tagged as awaiting-changes or awaiting-feedback by an @nf-core/modules contributor. Remove stale label or add a comment if it is still useful.

[sahuno]

@SPPearce thanks — renamed to modkit/extractcalls.

modkit/extract/calls was three levels deep, which fails the
main_nf_module_granularity lint check (<tool> or <tool>/<subtool> only).
Renamed the directory, the process name (MODKIT_EXTRACTCALLS), the meta.yml
name: field and the nf-test tags to match.

Sibling PR #11285 (modkit/extractfull) has had the same rename applied.

[SPPearce]

Generally fine now, please just tidy up the snapshot generation.

[sahuno]

@SPPearce all four addressed, thanks.

First a note that explains why the diff is bigger than the four comments: the
branch was ~980 commits behind master, so sanitizeOutput was not available on
it at all — it comes from nft-utils 1.x and the branch's nf-test.config
still pinned nft-utils@0.0.9. Rebased onto current master before making the
changes.

• Snapshot generation. Both tests now use your form: assert process.success
  outside assertAll, then a single snapshot(sanitizeOutput(process.out)).match().
  Snapshot regenerated — it drops the duplicated numbered keys and is about half
  the size.
• Redundant asserts. Removed from the non-stub test.
• .crai. You were right, and I checked rather than assumed: modkit extract calls reads CRAM. I built one from test.sorted.phased.bam against
  genome.fasta, indexed it, and got the same 27,813 rows and exit 0 as the BAM
  input. The index input is now pattern: "*.{bai,csi,crai}", and the bam
  description mentions that a CRAM input also needs fasta supplied.
• tests/nextflow.config. Deleted, along with the config "./nextflow.config"
  directive that loaded it — it only set ext.args = ''. The non-stub tsv md5 is
  unchanged (0ce2e5a6…), which confirms it was a no-op rather than something the
  test quietly depended on.

Both nf-tests pass locally under the singularity profile.

One thing to flag: lint now warns bioconda_latest: ont-modkit 0.6.1 -> 0.6.4.
I have left it at 0.6.1 to stay consistent with the merged modkit/pileup and my
three sibling modkit PRs (#11283, #11284, #11285). Happy to bump all of them
together in a follow-up if you would rather they move to 0.6.4.

[SPPearce]

At some point a bump to the newer version would be good, but they are better to be all on the same version as you say. Happy for a bump to occur within one of these new module PRs.

[SPPearce]

Ok, can merge this.
If you want, you can incorporate the version bump here.
Please update the other two modules with the same sanitizeOutput changes too (if you haven't already done this)

[sahuno]

Thanks @SPPearce. Both done:

• Version bump is in this PR — ff5d631 bumps ont-modkit 0.6.1 → 0.6.4 in environment.yml, and the snapshots were regenerated against it.
• The sibling modules already carry the same sanitizeOutput change — New module: modkit/extractfull #11285 (modkit/extractfull), New module: modkit/validate #11284 (modkit/validate), New module: modkit/entropy #11283 (modkit/entropy) and Add module: modkit/localize #11120 (modkit/localize) were each rebased onto master, switched to assert process.success + snapshot(sanitizeOutput(process.out)), had their snapshots regenerated and their containers bumped to 0.6.4. (modkit/entropy keeps its structural assertions on the non-stub tests — its float columns are not byte-deterministic across CPU counts — but the stubs use sanitizeOutput.)

The two review threads above are addressed and resolved. This one is green and mergeable whenever you are ready.