Conversation
* add ref data for pcgr 2.2.1 * add vep path pcgr * edit refdata path for vep pcgr * update pcgr report name name * change path VEP ref data * update container version * remove VEP version * change bolt version to test
* add process and config for vcf2maf * fix path vep data * remove duplicate * fix and linting * remove params from VCF2MAF process * fix and simplify process * add meta.key to channel * add genome build as input * add dockerfile * Make sigrap (HRDetect, MutationalPatterns) individual process (#16) * remove trace of genome build and hardcode in VCF2MAF * use biocontainers container for vcf2maf * remove string interpolation * change to stub no version template * revert deleted line
* change sigrap container * change cancer report container * latest bolt version * update meta id sigrap * bump sigrap * remove print * restaure metadata sigrap * change multiqc container * add param for pcgr chunck size * add mutpat predefined dbs argument * update container * add back formating comment
* fix channel prepare reference * refactor tarball extraction to use proper channel patterns. Follow oncoanalyser pattern for tarball extraction with .collect() and .map() * adapte extraction tarball to match pcgr expected vep structure * fix dir name extract tarball for VEP * change to adapt strip compoment for pcgr data
- Document optional dragen_somatic_vcf/dragen_germline_vcf samplesheet fields with example - Document pcgr_variant_chunk_size parameter for hypermutated samples - Update citation version from 0.6.0 to 0.7.0
…vent PURPLE receiving mnv_filtered intermediate
Ships the graceful PCGR skip on samples whose per-chunk variant count exceeds MAX_SOMATIC_VARIANTS, plus disable_estimates on chunked runs where MSI/TMB on partial VCFs is meaningless. Variant tag suffixes (-pcgr, -circos, -multiqc, -gpgr, -snpeff) preserved.
Moves hrdetect and mutpat modules off the personal docker.io/qclayssen dev build to the official umccr ghcr release image.
- prepare_input.nf: tumor_id/normal_id from explicit samplesheet columns;
sage path fallback sage_calling/ → sage/ (OA v2.3.0+); dragen germline
falls back to pave/{normal_id}.pave.germline.vcf.gz; dragen somatic
returns [] when absent (OA-only mode)
- sash.nf: branch somatic on dragen presence; sage somatic passes through
directly when no dragen (rescue skipped); germline steps gated on vcf
presence; somatic report passes [] for dragen vcf; MultiQC dragen dir
channels filtered when dirs absent
- bolt report module: container → 0.3.1-dev-oa; --vcf_dragen_fp conditional
- samplesheet.csv: add tumor_id/normal_id columns; add OA-only SBX example
groupByMeta requires >=2-element tuples (data[1..-1] throws on a single-element list). Passing ch_somatic_branch.oa_only directly emits [meta, [], []], which groupByMeta's recursive flatten collapses to [meta, sage_vcf, sage_tbi] — the correct 3-arg signature.
Drop DRAGEN somatic VCF, germline VCF, and HRD score handling entirely. Germline now always uses OA PAVE output; rescue step removed.
CPSR writes no output files when zero variants remain after classification filtering. Mark cpsr_report output optional and guard the mv command.
Process was removed from the OA-only workflow earlier; the module file was left behind. No includes or config selectors reference it.
Drop dragen_hrd from BOLT_OTHER_CANCER_REPORT input tuple and script — always empty in OA-only mode. Drop smlv_dragen_vcf input and dragen_arg from BOLT_SMLV_SOMATIC_REPORT — replaced by optional --vcf_dragen_fp guard already handled in bolt.
…mode - Build the PAVE germline VCF path from normal_id (matches the file's actual naming and the fixture NORMAL.pave.germline.vcf.gz) instead of tumor_id, which 404'd on every real run. - Require normal_id and the PAVE germline VCF like every other OA input, instead of silently degrading to tumor-only with no germline data. Tumor-only was never intended to be supported here, and the empty-VCF fallback caused samples to silently drop out of the MultiQC report entirely (fixed groupTuple join never completes).
ffc37ac (2026-07-13) changed this from tumor_id to normal_id, claiming tumor_id "404'd on every real run" - but that fix's own PR (#58) test plan left "End-to-end run against a real oncoanalyser output directory" unchecked, and the commit's justification ("matches ... the fixture NORMAL.pave.germline.vcf.gz") cites a fixture that same commit did not change, i.e. it matches its own synthetic fixture, not verified real output. Tested live against real production oncoanalyser 2.2.0 SEQC-II output (s3://test-data-503977275616-ap-southeast-2/testdata/analysis/ production/oncoanalyser-wgts-dna/2.2.0/SEQC-II/20260401d6eff920/): normal_id 404s, tumor_id resolves. Root-caused in oncoanalyser's own source (subworkflows/local/pave_annotation/main.nf): PAVE_GERMLINE's sample_id is set via Utils.getTumorDnaSampleName(meta) - oncoanalyser keys every per-pair output file, including the germline VCF, by the tumor sample ID, never the normal ID. Stable since 2024-05-02. Renamed the fixture (NORMAL -> TUMOR.pave.germline.vcf.gz) and updated the nf-test assertion to match, so the fixture now reflects real oncoanalyser output naming instead of the code's assumption. nf-test subworkflows/local/tests/prepare_input/main.nf.test: PASSED
4 tasks
This was referenced Jul 22, 2026
Conflicts resolved: - CHANGELOG.md: accept Breaking change note from main - cancer_report/main.nf: keep OA-only input tuple (no dragen_hrd) - smlv_somatic/rescue/main.nf: deleted (OA-only removes rescue) - prepare_input.nf: keep OA-only version (PAVE germline, no DRAGEN) - prepare_reference.nf: accept main's tarball-only policy - workflows/sash.nf: keep OA-only version (no rescue, no dragen_hrd) - tests/prepare_input: keep OA-only test suite
Resolves one real conflict: modules/local/bolt/smlv_somatic/rescue/main.nf was deliberately dropped as an orphan module on this branch (0cefd2b) while main independently bumped its container tag (3b9fb48) - kept the deletion since the module is unreferenced in workflows/sash.nf on this branch. Everything else auto-merged cleanly (bolt 0.3.0->0.3.1 container bumps, testing-validation docs/SOP additions from main).
Collaborator
Author
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
2 participants
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
dragen_somatic_dir,dragen_germline_dir,dragen_somatic_vcf,dragen_germline_vcf), theBOLT_SMLV_SOMATIC_RESCUEstep, and DRAGEN HRD score plumbing that depended on those inputs. DRAGEN support itself continues to live onrelease/0.7.0/main; this branch is scoped to the OA-only path.sage_calling/→sage/output directory rename (OA v2.3.0+).normal_idremains required — tumor-only analysis is out of scope for this mode.Depends on the companion bolt containers built from
oa-only-v2: umccr/bolt#34Known follow-ups (not blocking)
docs/usage.md,README.md, anddocs/details.mdstill document the removeddragen_*filetypes and DRAGEN HRD source — needs a doc pass in a follow-up PR.Test plan
nf-testforsubworkflows/local/prepare_input(updated to cover OA-only path resolution)Validation results
Both legs done — full tables posted on sash#62.
Note on TML/TMB in OA-only mode: OA-only germline calls come from PAVE (~1.9k
driver-focused variants), not DRAGEN's genome-wide germline VCF (~5M variants). This
shifts TML/TMB slightly vs a full DRAGEN+OA run on the same sample (SBJ04890: TML 75 vs
61, TMB/Mb 3.0598 vs 3.1123) — purity/ploidy/gender/QC/MSI are unaffected and match
exactly, since PURPLE's CN/SV calls come from the same OA-derived AMBER/COBALT input in
both modes. Don't read the TML/TMB difference as a regression.