Add dotseq/dotseq

[pinin4fjords]

Adds a dotseq/dotseq module wrapping DOTSeq, a Bioconductor package for detecting differential ORF usage (DOU) and ORF-level differential translation efficiency (DTE) from Ribo-seq with matched RNA-seq.

Status

• Bioconda recipe (bioconda/bioconda-recipes#65677) merged; bioconductor-dotseq=1.0.0 live on linux-64 + osx-64.
• Container: Wave community build pulling bioconductor-dotseq + tidyverse + plotting deps.
• Test data (nf-core/test-datasets#2072) open, awaiting maintainer review. The module's tests/nextflow.config points at the PR's fork branch until that merges.
• nf-core modules lint dotseq/dotseq clean.
• Local nf-test passes against the Wave container in ~4 min on AWS c5.9xlarge.

What it does

• Inputs: contrast tuple (variable, reference, target) + (samplesheet, count_matrix, annotation_tsv). The shape mirrors anota2seq/anota2seqrun and deltate (samplesheet + counts) with one extra per-ORF annotation TSV that supplies the parent-gene id the DOU beta-binomial fit needs.
  • samplesheet: CSV/TSV with run, strategy, replicate, condition columns (overridable via task.ext.args).
  • count_matrix: TSV with orf_id as the first column and per-sample count columns. Both Ribo-seq and RNA-seq samples share the matrix; the sample sheet's strategy column distinguishes them.
  • annotation_tsv: per-ORF rows with orf_id + gene_id (required), and optional orf_type (mORF/uORF/dORF) and coordinate columns. Genomic coordinates are stored on the resulting GRanges for downstream inspection only; the fit itself only uses gene_id (to group child ORFs per parent gene) and orf_type (to bucket the heatmap).
• Calls DOTSeqDataSetsFromSummarizeOverlaps() -> DOTSeq() -> getContrasts(). The R template uses optparse for --key value overrides via task.ext.args and tidyverse syntax throughout.
• Emits only what DOTSeq produces natively:
  • translation.dotseq.results.tsv - DTE (DESeq2 + ashr) interaction results
  • dou.dotseq.results.tsv - DOU (beta-binomial glmmTMB + ashr) interaction results
  • dou_strategy / dte_strategy TSVs when present (per-condition Ribo-vs-RNA contrasts)
  • volcano.png, composite.png, venn.png, heatmap.png from plotDOT()
  • interaction_p_distribution.png - histogram of DTE padj (plain ggplot on the package's own statistic)
  • DOTSeqDataSets.rds, R_sessionInfo.log, versions.yml

Why this input shape

The DOTSeq vignette uses DOTSeqDataSetsFromFeatureCounts() because the cell_cycle example was generated with featureCounts -f -O against a flattened ORF GTF; the constructor parses that featureCounts-format table and pairs it with a flattened GTF + BED to build the internal GRanges. That is not intrinsic to the DOTSeq model. The package also exports DOTSeqDataSetsFromSummarizeOverlaps(), which takes a plain per-ORF count matrix plus a GRanges annotation. The latter is a much cleaner contract for upstream consumers (nf-core/riboseq's ORF-level pipeline already produces a per-ORF P-site count matrix and a per-ORF annotation TSV directly, with no need to synthesize a featureCounts header + flattened GTF + flattened BED). The module wraps FromSummarizeOverlaps and builds the GRanges in-process from the annotation TSV.

Cross-tool naming

The DTE interaction table is written as translation.dotseq.results.tsv to match anota2seq/anota2seqrun's translation.anota2seq.results.tsv - same biological quantity (differential translation efficiency), measured per-ORF here vs per-gene there. This keeps downstream pipelines portable across the three differential translation methods (anota2seq, deltate, DOTSeq) while each module still emits only what its underlying package supports natively. DOTSeq's DOU output is unique to this tool and has no anota2seq/deltate counterpart.

Test

nextflow_process test using DOTSeq's own cell_cycle_subset bundled data (~150KB total, MIT licence, derived from inst/extdata), restricted to the Mitotic_Cycling vs Interphase contrast on chx-treatment samples. Snapshots filenames + versions topic + versions.yml md5; DOTSeq's Bayesian + glmmTMB outputs are stochastic so file content is not snapshotted.

PR checklist

• This comment contains a description of changes (with reason).
• Add a test for the new module/subworkflow.
• Snapshot the test output.
• If you've added a new tool - have you followed the module conventions in the contribution docs
• If necessary, include test data in your PR. (Sibling test-datasets PR linked above.)
• Remove all TODO statements.
• Emit the versions.yml file.