Add module: modkit/localize

[sahuno]

New module: modkit/localize

Tool

modkit — A bioinformatics tool for working with modified bases from Oxford Nanopore sequencing data.

What this module does

MODKIT_LOCALIZE aggregates bedMethyl pileup counts around genomic features of interest (e.g. CpG islands, gene bodies, repeat elements), producing:

• A TSV of percent-modification vs. offset from feature midpoints
• An optional interactive HTML chart

This enables investigation of modification enrichment or depletion patterns relative to genomic landmarks, analogous to a methylation profile around features.

Inputs

• bgzip-compressed + tabix-indexed bedMethyl file (from modkit pileup)
• Genome sizes file (e.g. .fai or .sizes)
• BED file of regions to localize around

Outputs

• *.tsv — percent-modification vs. offset table
• *.html — interactive chart
• *.log — debug log

Tests

Tests chain MODKIT_PILEUP → TABIX_TABIX → MODKIT_LOCALIZE using existing nf-core test-datasets (homo_sapiens nanopore BAM + genome.fasta.fai). Regions BED is created inline — no new test data upload required.

• nf-test passes locally (singularity profile)
• nf-test passes on SLURM compute node (job 17362918, 2 tests PASSED in 13s)
• nf-core modules lint passes: 48/48 tests, 0 warnings, 0 failures

Notes

• --regions is required by the modkit 0.6.1 CLI (enforced at the tool level)
• Container spec matches the existing modkit/pileup module
• A companion modkit/localize/plot submodule for composite multi-sample visualization is planned as a follow-up PR

[famosab]

We usually recommend to have one module per PR. That makes the review process easier and it is more likely that someone will review your PR. I would suggest to split this PR up into the dorado module and the modkit module :)

[sahuno]

@famosab you were right, and sorry this took so long. Split done — this PR is
now modkit/localize only, six files, nothing else.

• dorado/basecaller is gone from here entirely. It is already Add module: dorado/basecaller #11122, and the
  copy in this branch was the older revision that still committed pod5 files
  into the repo.
• modkit/localize/plot moves to its own module, renamed modkit/localizeplot
  (three-level paths fail main_nf_module_granularity). I have not opened that
  PR yet — see the note at the bottom.

Rebasing onto current master surfaced two things that were quietly broken:

TABIX_TABIX is deprecated. Its script block on master is now
assert false with a message pointing at HTSLIB/BGZIPTABIX, and its input
signature had grown from two elements to four. The test setup now uses
HTSLIB_BGZIPTABIX to build the tabix index from the MODKIT_PILEUP output.

The test config was never loaded. main.nf.test had no config directive,
so params.module_args never reached ext.args and the
--min-coverage 1 --window 1000 values the test depends on were being silently
ignored. Added the directive.

Also aligned the container declaration with the merged modkit modules
(workflow.containerEngine in ['singularity', 'apptainer'], quay.io/ prefix).

nf-core/tools 4.1.0 lint: 59 passed, 0 warnings, 0 failed. Both nf-tests
pass under the singularity profile.

On modkit/localizeplot: it is ready except for its container, which currently
points at a temporary Wave build URI that is neither on the allowed registry
list nor stable. It needs a frozen community.wave.seqera.io/library/... image
built from its environment.yml. I would rather sort that out before opening the
PR than add another red one to the queue — but if you would prefer to see it now
regardless, say the word and I will open it.

[sahuno]

Update since the split: rebased onto master, tests moved to assert process.success + snapshot(sanitizeOutput(process.out)) (same pattern as #11286), snapshots regenerated, container bumped to ont-modkit 0.6.4. This PR is modkit/localize only — six files — CI is green and there are no open review threads. @famosab re-requesting review since you flagged the original bundle.