Add DOTSeq cell_cycle subset test data for modules#2072
Open
pinin4fjords wants to merge 5 commits into
Open
Conversation
Bundled inst/extdata files from the DOTSeq Bioconductor package, copied into homo_sapiens/riboseq_expression/dotseq for use by the nf-core/modules dotseq/dotseq nf-test: - featureCounts.cell_cycle_subset.txt.gz: ORF-level featureCounts output - gencode.v47.orf_flattened_subset.gtf.gz: flattened ORF annotation - gencode.v47.orf_flattened_subset.bed.gz: matching BED - metadata.txt.gz: sample condition table (run/strategy/replicate/treatment/condition) Source: https://github.com/compgenom/DOTSeq, MIT license.
Derived from metadata.txt.gz, filtered to treatment=chx samples (which is what the bundled featureCounts table contains) and headered as run,strategy,replicate,condition for direct consumption by the dotseq/dotseq nf-test.
7 tasks
Member
Author
|
Companion to nf-core/modules#11742 - the dotseq module nf-test 404s on this samplesheet URL until this PR merges, otherwise green. |
Document the file set, derivation, and verification of the DOTSeq cell_cycle_subset fixtures, matching the gedi/price test data PR convention.
pinin4fjords
added a commit
to pinin4fjords/nf-core-modules
that referenced
this pull request
May 22, 2026
Lets CI verify the module is actually green; revert this commit once nf-core/test-datasets#2072 merges and the canonical modules-branch URL resolves.
Replaces the upstream featureCounts table and flattened GTF/BED with a tidy per-ORF count matrix (counts.tsv.gz) and per-ORF annotation TSV (annotation.tsv.gz) so the dotseq module can call DOTSeqDataSetsFromSummarizeOverlaps() directly. ORF identifiers and sample columns are preserved; both files are derived from the same DOTSeq cell_cycle_subset cohort. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
pinin4fjords
added a commit
to pinin4fjords/nf-core-modules
that referenced
this pull request
May 22, 2026
Aligns the module's input contract with deltate / anota2seq so that consumers can dispatch between the three ORF-DTE methods without maintaining a separate prep step for dotseq. The four featureCounts/GTF/BED inputs collapse to a per-ORF count matrix (orf_id + sample columns) plus a per-ORF annotation TSV (orf_id + gene_id + optional orf_type/coords). The R template now calls DOTSeqDataSetsFromSummarizeOverlaps() and builds the required GRanges in-process from the annotation TSV; the model fit, contrast tables, and plotDOT outputs are unchanged. Test fixtures updated alongside in nf-core/test-datasets#2072 (commit 8c9b27c). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
toniher
pushed a commit
to biocorecrg/nf-core-modules
that referenced
this pull request
Jun 12, 2026
* Add dotseq/dotseq module DOTSeq is a Bioconductor package for detecting differential ORF usage (DOU) and ORF-level differential translation efficiency (DTE) from Ribo-seq with matched RNA-seq. Module wraps DOTSeqDataSetsFromFeatureCounts + DOTSeq() + getContrasts() and emits per-ORF TSVs for the DOU and DTE interaction contrasts plus the serialised DOTSeqDataSets object. Pre-requisites (in flight): - Bioconda recipe: bioconda/bioconda-recipes#65677 - Test data: nf-core/test-datasets#2072 * Use Wave community container for bioconductor-dotseq 1.0.0 Bioconda recipe (bioconda/bioconda-recipes#65677) merged; biocontainer image is not yet built so swap the placeholder quay.io/depot URLs for a Wave community container built from the now-merged bioconda package. Also widen the singularity guard to include 'apptainer' and add the versions topic block in meta.yml (via nf-core modules lint --fix). * Add native plotDOT outputs, simplify template, tidyverse syntax - Restructure the R template around optparse + readr + dplyr + purrr + ggplot2; drop the homemade parse_args / read_delim_flexible helpers in favour of the standard package idioms and native pipe. - Output set is now what DOTSeq itself emits natively: per-ORF DTE contrasts (translation.dotseq.results.tsv), DOU contrasts (dou.dotseq.results.tsv), optional dou_strategy / dte_strategy per-condition Ribo-vs-RNA contrasts, plus the four plotDOT() PNGs (volcano / composite / venn / heatmap) and a DTE p-value distribution histogram drawn directly from DOTSeq's padj column. - Container picks up r-eulerr + r-ggsignif (required for plotDOT venn) and explicit r-ggplot2 so the histogram has a stable ggplot version. - plotDOT() default of force_new_device=TRUE was killing our png() device on each call; pass FALSE so the PNGs land where Nextflow expects them. * Simplify R template helpers, add heatmap sorf_type fallback - Drop the homemade read_delim_flexible() and write_results_tsv() wrappers in favour of read_tsv() / read_csv() / write_tsv() directly. The earlier to_orf_tibble() conditional is also gone now that we know getContrasts() always returns a frame with orf_id as a column (per the DOTSeq source in posthoc.R + main.R). - plotDOT(heatmap) requires gene-paired mORF + sorf entries; try uORF first (the package default) and fall back to dORF when no significant gene has both. tryCatch in safe_plot_dot still makes either a no-op when neither succeeds. * Address code-review feedback: stub block, validation hardening, plot fallback robustness - Add stub: block to main.nf matching the proteus/readproteingroups precedent. - Read sample sheet with read_delim() picking comma/tab from the file extension so the meta.yml-advertised TSV variant actually works. - Refuse to clobber an existing canonical column (e.g. an existing 'condition' column when --contrast_variable=treatment is supplied). - Dedupe multi-lane sample sheets and validate that both Ribo and RNA strategies are present (DOTSeq's interaction design is unestimable otherwise). - Add an is_set() predicate that catches NULL / empty stringent + required options before the tri-state switch silently returns NULL. - safe_plot_dot now unlinks the partially-written PNG on plotDOT error and returns success so the heatmap fallback (uORF then dORF) keys off whether the first call actually drew, not file.exists() of a stale handle. - getContrasts(type='interaction') errors propagate (headline outputs); type='strategy' stays tryCatch'd because absence is legitimate. - Cache getDOU(d) / getDTE(d) once and share across contrasts + plotDOT. - Drop redundant file.exists() walk - Nextflow's path staging already guarantees the inputs exist. - Expand the test to assert volcano / composite / venn plot emission and add a -stub test. * TEMPORARY: point test at the pending test-datasets PR fork branch Lets CI verify the module is actually green; revert this commit once nf-core/test-datasets#2072 merges and the canonical modules-branch URL resolves. * refactor(dotseq/dotseq): take a count-matrix shape for consumer parity Aligns the module's input contract with deltate / anota2seq so that consumers can dispatch between the three ORF-DTE methods without maintaining a separate prep step for dotseq. The four featureCounts/GTF/BED inputs collapse to a per-ORF count matrix (orf_id + sample columns) plus a per-ORF annotation TSV (orf_id + gene_id + optional orf_type/coords). The R template now calls DOTSeqDataSetsFromSummarizeOverlaps() and builds the required GRanges in-process from the annotation TSV; the model fit, contrast tables, and plotDOT outputs are unchanged. Test fixtures updated alongside in nf-core/test-datasets#2072 (commit 8c9b27c). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(dotseq/dotseq): synthesize `replicate` column when absent DOTSeq's parse_condition_table() requires a `replicate` column for stable ordering of samples within strategy+condition. Pipeline samplesheets often have a `pair` column (or none at all), so the R template now treats the column as optional: when present it is renamed to `replicate` as before; when absent the template assigns a per-(strategy, condition) row counter so the model fit is unaffected. This matches how anota2seq/deltate consume the same samplesheets. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(dotseq/dotseq): support running a single module Dropping a module from --modules left DOTSeq()'s skipped slot unfitted (a bare DESeqDataSet for DTE), and getContrasts() has no method for it, so a DOU-only run crashed when extracting the DTE interaction table. Gate interaction and strategy contrast extraction on the selected modules, and write each module's interaction table only when that module ran. Mark the translation and dou outputs optional to match, and add a DOU-only regression test. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Test fixtures for nf-core/modules#11742 (
dotseq/dotseq) underdata/genomics/homo_sapiens/riboseq_expression/dotseq/.DOTSeq is a Bioconductor package for differential ORF usage (DOU) + ORF-level differential translation efficiency (DTE). Its API takes a featureCounts-format ORF count table plus a flattened ORF GTF/BED pair, so the existing
riboseq_expression/salmon-gene-level fixtures don't satisfy the input contract. Rather than build new featureCounts + ORF annotations from the existing BAMs (which would require an ORF caller + featureCounts pass), this PR drops in the smallcell_cycle_subsetbundled with the DOTSeq package - DOTSeq's own example data, already in the exact shape its functions expect.Files
featureCounts.cell_cycle_subset.txt.gzgencode.v47.orf_flattened_subset.gtf.gzgencode.v47.orf_flattened_subset.bed.gzmetadata.txt.gzrun strategy replicate treatment conditionsamplesheet.csvmetadata.txt.gz(12 rows) - derived in this PR so the module can consume it without an extra header stepTotal: 252 KB across 5 files.
Why these specific files
DOTSeq's
DOTSeqDataSetsFromFeatureCounts()requires four inputs:count_tablewith featureCounts annotation columns (Geneid, Chr, Start, End, Strand, Length) followed by per-sample counts.flattened_gtfwithgene_id+exon_numberattributes (DOTSeq uses these to name ORFs asgene_id:O###).flattened_bedmatching the GTF.condition_tablewithrun,strategy,replicate,conditioncolumns.The first four files are reused verbatim from DOTSeq's
inst/extdata/.samplesheet.csvis the only derived file:metadata.txt.gzships as a 5-column whitespace table with no header, so the module test would need to add a header before consuming it. To keep the nf-test minimal we ship the headered, chx-filtered version (12 rows matching the 12 sample columns in the featureCounts file) directly.Source & licence
Bundled
inst/extdataof compgenom/DOTSeq (Bioconductor 3.23 release). Upstream data is from Ly et al. 2024 (GSE231096 / SRR242304XX cell-cycle Ribo-seq + RNA-seq cohort), captured as the DOTSeq author's example dataset. Package licence: MIT.Test plan
modules/nf-core/dotseq/dotseq/tests/main.nf.testin nf-core/modules#11742 consumes these fixtures viaraw.githubusercontent.comURLs pinned to this PR's branch; URLs to be updated tomodulesafter this PR mergesnf-core modules test --profile docker dotseq/dotseqruns DOTSeq's DOU + DTE + per-strategy DESeq2 fits + plots end-to-end in ~4 min on a c5.9xlarge against this fixture set