Skip to content
Report an issue
Pipelines / oxo-flow-varlociraptor

Small and structural variant calling with Varlociraptor

✔ Live-tested Official port snakemake portvariant-callingstructural-variantsvarlociraptorvgfreebayesdellypangenomesnakemake-workflows

Scenario-driven somatic small and structural variant calling with Varlociraptor: paired-end reads are aligned against the 1000 Genomes human pangenome with vg giraffe, QC'd with FastQC/MultiQC, covered with mosdepth, and used for freebayes and delly candidate calling; Varlociraptor then estimates alignment properties and calls variants under a tumor scenario (events present + somatic_tumor_high + somatic_tumor_medium, FDR 0.05), FDR is controlled per variant type (SNV/INS/DEL/MNV/BND/INV/DUP/REP) with merge and phred decoding, and the calls are annotated with VEP (LoFtool/REVEL plugins) and dbSNFP/dbSNP, filtered, turned into a 34-column variant table with oncoprint label-sorting, and rendered as interactive datavzrd variant and gene-coverage reports. All reference data (GRCh38 FASTA and GTF, VEP cache/plugins, REVEL scores, known-variants VCFs, HPRC pangenome graph) is downloaded automatically into resources/.

▶ Run itGitHub ↗$ oxo-flow run main.oxoflow
At a glance
Rating✔ Live-tested
Rules149
Computeup to 64 CPUs / 32 GB per rule (freebayes candidates 48 threa…
Enginesnakemake port
OriginOfficial port
Domaingenomics
Pinned versionv6.10.0
Ported2026-08-15
LicenseApache-2.0
Tools
altairarribabcftoolsbedtoolsbiopythonbwacurldatavzrd

$ oxo-flow pull gh:oxo-flow-community/oxo-flow-varlociraptor

Semantic overview — plain-language walkthrough text

VarLociRaptor small-and-structural variant calling: one tumor sample's reads are pangenome-mapped, candidates called with freebayes and delly, re-filtered by Varlociraptor scenario, FDR-controlled per event type, and delivered as annotated calls plus interactive reports.

1. Reference and read preparation — get_genome/get_annotation fetch the genome and annotation, feeding genome indexes (genome_faidx, genome_dict), the pangenome (get_pangenome → pangenome_autoindex → get_reference_paths), VEP cache/plugins (get_vep_cache, get_vep_plugins) and REVEL scores (download_revel → process_revel_scores → tabix_revel). Reads enter alignment through merged trimmed FASTQs (merge_trimmed_fastqs_r1/merge_trimmed_fastqs_r2), with fastp_pe/fastp_se/fastp_pipe trimming optional.

2. Alignment, BQSR and QC — map_reads_vg aligns to the pangenome, then postprocess_vg_alignments → sort_alignments → mark_duplicates → bam_index_dedup → recalibrate_base_qualities → apply_bqsr yields the calibrated BAM; multiqc gathers fastqc_r1/fastqc_r2, samtools_stats/samtools_idxstats. A gated consensus branch (calc_consensus_reads → map_consensus_reads_pe/map_consensus_reads_se → apply_bqsr_consensus) reconverges downstream.

3. Regions and primers — build_sample_regions computes per-base coverage; merge_expanded_group_regions → filter_group_regions_expanded and merge_covered_group_regions → filter_group_regions_covered restrict candidate calling. The primer chain (map_primers → filter_unmapped_primers → primer_to_bed → build_primer_regions → assign_primers → filter_primerless_reads → trim_primers) filters reads by primer.

4. Candidates and Varlociraptor — freebayes/delly call candidates (delly also uses download_delly_excluded_regions); fix_delly_calls cleans delly output; scatter_candidates_freebayes/scatter_candidates_delly split callsets for annotate_candidate_variants_freebayes/annotate_candidate_variants_delly (VEP) and filter_candidates_by_annotation_freebayes/filter_candidates_by_annotation_delly. After varlociraptor_alignment_properties, the varlociraptor_preprocess_freebayes/varlociraptor_preprocess_delly observations feed varlociraptor_call_freebayes/varlociraptor_call_delly under the render_scenario scenario, converging at bcftools_concat.

5. Annotation, FDR control and reporting — bcftools_concat output is VEP-annotated (annotate_variants → bcf_index_vep_annotated → annotate_vcfs, optional annotate_dgidb); after filter_by_annotation and gather_calls, per-type FDR control (control_fdr_SNV, control_fdr_INS, control_fdr_DEL, control_fdr_MNV, control_fdr_BND, control_fdr_INV, control_fdr_DUP, control_fdr_REP) leads to merge_calls → convert_phred_scores → vembrane_table → process_call_tables, alongside prepare_oncoprint → datavzrd_variants_calls and bedtools_merge → coverage_table → datavzrd_coverage.

6. Gated branches — population-DB filtering (gather_annotated_calls, population_filter_variants), MAF export (group_bcf_to_vcf_variants, group_vcf_to_maf_variants), COSMIC signatures (create_mutational_context_file, annotate_mutational_signatures, plot_mutational_signatures), fusion calling (star_index, star_align, arriba, sort_arriba_calls) and CHM benchmarking (chm_eval).

Verified: every rule name above is a real rule of main.oxoflow (oxo-flow validate); the described order follows the actual rule dependencies.

Report a correction to this overview

Run it

oxo-flow run gh:oxo-flow-community/oxo-flow-varlociraptor

Runs straight from the catalog — oxo-flow checks the repo out under .oxo-flow/repos/oxo-flow-varlociraptor and keeps outputs/checkpoints in the current directory, no manual clone. Pin a revision with gh:oxo-flow-community/oxo-flow-varlociraptor@<branch-or-tag>.

Preview the plan first: oxo-flow pull gh:oxo-flow-community/oxo-flow-varlociraptor fetches the repo, then oxo-flow dry-run main.oxoflow.

Needs reference data — see Requirements; preview with oxo-flow dry-run main.oxoflow --samples first:1.

Installation

Engine. oxo-flow >= 0.12.0

Toolchain. conda envs — pinned versions (no containers)

Requirements.

  • paired-end FASTQ reads at reads_dir/_1.fastq.gz / _2.fastq.gz; sample cohort declared in [[sample_groups]] (one group = one tumor sample); fixtures bundled for dry-run
  • reference data: downloaded automatically into resources/ — GRCh38 primary assembly FASTA (Ensembl release 111) + .fai/.dict, Ensembl release 111 GTF, VEP cache and plugins (release 111), REVEL scores, Ensembl known-variants VCFs, HPRC v1.1 human pangenome graph
  • compute: up to 64 CPUs / 32 GB per rule (freebayes candidates 48 threads — upstream 96, scaled; vg giraffe 64 threads; samtools sort 16 threads/32G; Varlociraptor call 8G; consensus/bam-name sorting 16 threads/64G when the gated branches are on)
  • tools: conda envs with pinned versions (envs/*.yaml, one env per tool pin set); conda/mamba required at runtime
  • disk: multi-GB reference downloads under resources/ (pangenome graph, VEP cache, known-variants VCFs) plus results/ for BAMs, BCFs, tables and reports
# 1. install oxo-flow (release binary, recommended)
curl -fL -o oxo-flow.tar.gz https://github.com/Traitome/oxo-flow/releases/latest/download/oxo-flow-latest-x86_64-unknown-linux-gnu.tar.gz
tar xzf oxo-flow.tar.gz && sudo mv oxo-flow /usr/local/bin/
#    or, via conda (may lag behind releases):
#    conda install -c bioconda oxo-flow-cli
#    NOTE: bioconda currently ships 0.10.2, older than the >= 0.12.0
#    minimum of every catalog entry — prefer the release binary.

# 2. get this workflow (clones the repo, auto-discovers the workflow,
#    sanity-parses it with the engine)
oxo-flow pull gh:oxo-flow-community/oxo-flow-varlociraptor
#    (alternative: plain git clone)
#    git clone https://github.com/oxo-flow-community/oxo-flow-varlociraptor

Parameters

Parameters are consumed by rules through {config.key} placeholders in inputs, outputs, and shells. Set a value in the workflow's [config] section (edit the file), or override at run time with oxo-flow run -e key=value workflow.oxoflow — repeat -e for multiple keys. Copy a row to paste the key directly. Click any parameter name to copy key = value; clicking default copies just the value.

ParameterTypeDefaultDescription
string db_annotated The annotated-callset selection for gather_annotated_calls (upstream
get_final_selected_annotation): "db_annotated" (annotations/vcfs active,
the default), "dgidb_annotated" when dgidb is activated, or "vep_annotated"
when annotations/vcfs is deactivated.
used by 3 rules
bool false upstream config: benchmarking (Snakefile rule benchmark / benchmarking.smk;
the full CHM-eval flow is ported, except the chm sample group vertical
slice (see module header)).
used by 8 rules
bool false upstream config: linear-reference (bwa) aligner branch of mapping.smk
(map_reads_bwa + ref.smk bwa_index). The default path aligns with vg
giraffe to the pangenome (ref/pangenome/activate = true upstream).
used by 2 rules
string GRCh38 upstream config: plugins (download_cadd_scores_for_vep; download_revel and
process_revel_scores are already ported in ref.oxoflow). cadd_build /
cadd_version / cadd_variant_type are the upstream wildcards of the same
rule with their upstream defaults.
used by 1 rules
string snv upstream config: plugins (download_cadd_scores_for_vep; download_revel and
process_revel_scores are already ported in ref.oxoflow). cadd_build /
cadd_version / cadd_variant_type are the upstream wildcards of the same
rule with their upstream defaults.
used by 1 rules
string v1.7 upstream config: plugins (download_cadd_scores_for_vep; download_revel and
process_revel_scores are already ported in ref.oxoflow). cadd_build /
cadd_version / cadd_variant_type are the upstream wildcards of the same
rule with their upstream defaults.
used by 1 rules
bool false upstream config: calc_consensus_reads/activate (consensus.oxoflow: rbt
collapse-reads-to-fragments + re-mapping to the linear bwa reference).
Activating it also needs the bwa reference index (bwa_align_activate or
primers_activate builds it) and, upstream-faithful,
markduplicates_extra = "--TAG_DUPLICATE_SET_MEMBERS true" and
freebayes_min_alternate_count = 1 (see those keys below).
used by 10 rules
bool false upstream config: annotations/dgidb (annotate_dgidb; datasources = [DrugBank]).
Activating dgidb changes which annotated callset the final-calls chain
consumes upstream (get_final_selected_annotation): set annotation_selection
below to "dgidb_annotated" together with this key.
used by 1 rules
int 2 upstream config: params/freebayes — the min-alternate-count for candidate
calling (2, or 1 when calc_consensus_reads is active).
used by 1 rules
bool false upstream config: fusion calling branch (fusion_calling.smk star_arriba
meta wrapper: star_index / star_align / arriba / annotate_exons /
convert_fusions / sort_arriba_calls / bcftools_concat_candidates).
used by 8 rules
bool false upstream config: maf/activate (group_bcf_to_vcf + group_vcf_to_maf).
used by 4 rules
string upstream config: params/picard/MarkDuplicates + get_markduplicates_extra —
extra MarkDuplicates arguments (upstream adds
"--TAG_DUPLICATE_SET_MEMBERS true" when calc_consensus_reads is active;
empty by default).
used by 1 rules
bool false upstream config: mutational_burden/activate + events
(calculate_covered_coding_sites + estimate_mutational_burden; events are
comma-joined, split to space-separated in the rule shells).
used by 6 rules
string somatic_tumor_low,somatic_tumor_medium,somatic_tumor_high upstream config: mutational_burden/activate + events
(calculate_covered_coding_sites + estimate_mutational_burden; events are
comma-joined, split to space-separated in the rule shells).
used by 2 rules
bool false upstream config: mutational_signatures/activate
(create_mutational_context_file ... plot_mutational_signatures; upstream
default events = [some_id], samples = [tumor], frozen in the module).
used by 8 rules
bool false upstream config: plugins (download_cadd_scores_for_vep; download_revel and
process_revel_scores are already ported in ref.oxoflow). cadd_build /
cadd_version / cadd_variant_type are the upstream wildcards of the same
rule with their upstream defaults.
used by 1 rules
bool false upstream config: population/db/activate + path/alias/fdr/events
(rules clean_population_db / population_filter_variants / population_db_update).
used by 7 rules
string tumor upstream config: population/db/activate + path/alias/fdr/events
(rules clean_population_db / population_filter_variants / population_db_update).
used by 1 rules
string somatic_tumor_high,somatic_tumor_medium upstream config: population/db/activate + path/alias/fdr/events
(rules clean_population_db / population_filter_variants / population_db_update).
used by 1 rules
float 0.05 upstream config: population/db/activate + path/alias/fdr/events
(rules clean_population_db / population_filter_variants / population_db_update).
used by 1 rules
string resources/population_db.variants.bcf upstream config: population/db/activate + path/alias/fdr/events
(rules clean_population_db / population_filter_variants / population_db_update).
used by 2 rules
bool false upstream config: primers/trimming (rules assign_primers ... build_primer_regions).
primers_fa1/primers_fa2 are the upstream primers/trimming/primers_fa{1,2} fasta
files (empty upstream default = primers flow off); fa2 empty means single-end
primer fasta.
used by 8 rules
string upstream config: primers/trimming (rules assign_primers ... build_primer_regions).
primers_fa1/primers_fa2 are the upstream primers/trimming/primers_fa{1,2} fasta
files (empty upstream default = primers flow off); fa2 empty means single-end
primer fasta.
used by 1 rules
string upstream config: primers/trimming (rules assign_primers ... build_primer_regions).
primers_fa1/primers_fa2 are the upstream primers/trimming/primers_fa{1,2} fasta
files (empty upstream default = primers flow off); fa2 empty means single-end
primer fasta.
not referenced by any rule (ported for upstream compatibility — overriding has no effect here)
string test/fixtures/raw Path to the raw paired-end FASTQs of the single sample (upstream
config/units.tsv points at absolute /projects/... paths; the port reads
from the repository fixtures instead).
used by 7 rules
bool false Skip the ref:: download rules (genome, annotation, VEP cache/plugins,
pangenome, REVEL, known variants — ~5 GB of public databases). Set to
true when you have pre-placed the files at the resource paths the rules
declare (see README "Reference databases"); the downloads are hardcoded
upstream URLs and need unimpeded network access.
used by 7 rules
bool false upstream config: trimming (get_sra / fastp rules). The default path has no
trimming configured — reads pass through mapping::merge_trimmed_fastqs.
used by 4 rules

Descriptions are the workflow's own # comments from its [config] section (and the [config] sections of its included modules), surfaced by oxo-flow info — no schema file to maintain.

Workflow graph

Overview — all modules
oxo-flow-varlociraptor pipeline overview

figure · oxo-flow-varlociraptor — Scenario-driven somatic small and structural variant calling with Varlociraptor: paired-end reads are aligned against the 1000 Genomes human pangenome with vg giraffe, QC'd with FastQC/MultiQC, covered with mosdepth, and used for freebayes and delly candidate calling; Varlociraptor then estimates alignment properties and calls variants under a tumor scenario (events present + somatic_tumor_high + somatic_tumor_medium, FDR 0.05), FDR is controlled per variant type (SNV/INS/DEL/MNV/BND/INV/DUP/REP) with merge and phred decoding, and the calls are annotated with VEP (LoFtool/REVEL plugins) and dbSNFP/dbSNP, filtered, turned into a 34-column variant table with oncoprint label-sorting, and rendered as interactive datavzrd variant and gene-coverage reports.

Read: stations are rules (or module groups); a line is a data dependency; stations without any line are off-track inputs/terminal exports with no dataflow edge; separate groups of lines are independent chains (e.g. a quantifier reading raw reads while the alignment chain runs aside — live: tcasia salmon_quant). The map shows the template DAG; oxo-flow graph --expanded adds one node per sample instance.

The graph is derived at catalog-build time from oxo-flow graph -f metro through the adaptive render ladder (scripts/metro_tiers.py): each workflow gets the finest metro tier that nf-metro renders while staying readable at site width — rule-level stations for smaller workflows, module-stage or moduleoverview stations for dense ones. Colored transit lines group stations by analysis stage. Wildcard {sample} instances expand at run time when sample data is discovered (the runtime view is oxo-flow graph --expanded).

Scope

The default-parameters main path of the source pipeline was ported rule-for-rule; alternate paths are documented as excluded.

In scope

  • annotate_candidate_variants_delly
  • annotate_candidate_variants_freebayes
  • annotate_variants
  • annotate_vcfs
  • apply_bqsr
  • bam_index_dedup
  • bcf_index_arriba
  • bcf_index_candidate_delly
  • bcf_index_candidate_freebayes
  • bcf_index_delly
  • bcf_index_fdr_BND
  • bcf_index_fdr_DEL
  • bcf_index_fdr_DUP
  • bcf_index_fdr_INS
  • bcf_index_fdr_INV
  • bcf_index_fdr_MNV
  • bcf_index_fdr_REP
  • bcf_index_fdr_SNV
  • bcf_index_filtered
  • bcf_index_freebayes
  • bcf_index_fusions_callset
  • bcf_index_vep_annotated
  • bcftools_concat
  • bcftools_concat_fusions
  • bcf_index_fdr_BND_fusions
  • bedtools_merge
  • build_sample_regions
  • control_fdr_BND
  • control_fdr_BND_fusions
  • control_fdr_DEL
  • control_fdr_DUP
  • control_fdr_INS
  • control_fdr_INV
  • control_fdr_MNV
  • control_fdr_REP
  • control_fdr_SNV
  • convert_phred_scores
  • convert_phred_scores_fusions
  • coverage_table
  • datavzrd_coverage
  • datavzrd_variants_calls
  • delly
  • download_delly_excluded_regions
  • download_revel
  • fastqc_r1
  • fastqc_r2
  • filter_by_annotation
  • filter_candidates_by_annotation_delly
  • filter_candidates_by_annotation_freebayes
  • filter_group_regions_covered
  • filter_group_regions_expanded
  • filter_offtarget_variants_delly
  • filter_offtarget_variants_freebayes
  • fix_delly_calls
  • freebayes
  • gather_calls
  • gather_annotated_calls_fusions
  • genome_dict
  • genome_faidx
  • get_annotation
  • get_genome
  • get_known_variants
  • get_pangenome
  • get_reference_paths
  • get_target_regions
  • get_vep_cache
  • get_vep_plugins
  • map_reads_vg
  • mark_duplicates
  • merge_calls
  • merge_calls_fusions
  • merge_covered_group_regions
  • merge_expanded_group_regions
  • merge_trimmed_fastqs_r1
  • merge_trimmed_fastqs_r2
  • multiqc
  • pangenome_autoindex
  • postprocess_vg_alignments
  • prepare_oncoprint
  • process_call_tables
  • process_revel_scores
  • recalibrate_base_qualities
  • remove_iupac_codes
  • render_scenario
  • samtools_idxstats
  • samtools_stats
  • scatter_candidates_delly
  • scatter_candidates_freebayes
  • sort_alignments
  • sort_calls_arriba
  • sort_calls_delly
  • sort_calls_freebayes
  • tabix_noiupac
  • tabix_revel
  • tabix_variation
  • transform_gene_annotations
  • varlociraptor_alignment_properties
  • varlociraptor_call_arriba
  • varlociraptor_call_delly
  • varlociraptor_call_freebayes
  • varlociraptor_preprocess_arriba
  • varlociraptor_preprocess_delly
  • varlociraptor_preprocess_freebayes
  • vembrane_table
  • annotate_dgidb
  • gather_benchmark_calls
  • chm_eval_sample
  • chm_namesort
  • chm_to_fastq
  • chm_eval_kit
  • chromosome_map
  • rename_chromosomes
  • chm_eval
  • determine_coding_regions
  • bcf_index_final
  • calculate_covered_coding_sites
  • estimate_mutational_burden_curve
  • estimate_mutational_burden_hist
  • create_mutational_context_file
  • download_cosmic_signatures
  • annotate_mutational_signatures
  • join_mutational_signatures
  • annotate_descriptions
  • plot_mutational_signatures
  • calc_consensus_reads
  • map_consensus_reads_pe
  • map_consensus_reads_se
  • merge_consensus_reads
  • sort_consensus_reads
  • bam_index_consensus
  • recalibrate_base_qualities_consensus
  • apply_bqsr_consensus
  • star_index
  • star_align
  • arriba
  • annotate_exons
  • convert_fusions
  • sort_arriba_calls
  • bcftools_concat_candidates
  • group_bcf_to_vcf_variants
  • group_bcf_to_vcf_fusions
  • group_vcf_to_maf_variants
  • group_vcf_to_maf_fusions
  • bwa_index
  • map_reads_bwa
  • download_cadd_scores_for_vep
  • clean_population_db
  • gather_annotated_calls
  • annotated_index
  • bcf_index_cleaned_db
  • population_filter_variants
  • bcf_index_population_filtered
  • population_db_update
  • assign_primers
  • filter_primerless_reads
  • trim_primers
  • map_primers
  • filter_unmapped_primers
  • primer_to_bed
  • build_primer_regions
  • get_sra
  • fastp_pipe
  • fastp_se
  • fastp_pe
  • bcf_index_arriba_candidates

Excluded

  • fusions report and table exports (upstream datavzrd.smk: datavzrd_fusion_calls + process_fusion_call_tables; and the fusions instance of the single wildcard vembrane_table rule) — the ported fusions FDR-control chain (filtering.oxoflow) ends at the fdr-controlled/normal-probs callsets; the port's vembrane_table hardcodes the variants path, so the fusions table/report datasets are missing. fusion_activate is the port's config key (upstream gates on the types: ["variants","fusions"] list). Note: upstream has no fusions oncoprint output — get_oncoprint_input is variants-only, so nothing to port there
  • annotate_umis (upstream mapping.smk:77, UMI branch via umi_tools group; activates for samples with a non-empty umi_read column) and splitncigarreads (upstream mapping.smk:173, RNA-datatype branch in get_recalibrate_quality_input) — not ported; the port models DNA data with the default empty samples columns
  • upstream testcase.smk debug module (gather_observations + testcase: varlociraptor call --testcase-prefix/--testcase-locus mode, results/testcases/) — not ported. (The upstream bcf_to_vcf_gz, vg2svg and only_alignment tools have no consumers/producers in the upstream tree at the pinned tag — dead tools, intentionally not ported)

Fidelity

The port aims for byte-identical commands on the default path. Known, deliberate deviations:

upstream port reason
scatter.calling(16) (rules run 16x, once per scatter item) single chunk, scatteritem=0 the port freezes scatteritem=0; rbt vcf-split with one output chunk writes the whole callset, so the chunk content is identical to upstream's 16 chunks gathered with bcftools concat -a before control_fdr (oxo-flow does have a scatter construct; it is not exercised because the port's single sample makes the split work-identical)
rule outputs that are directories (VEP cache/plugins, oncoprint label_sortings//variant-oncoprints/ dirs) directory + .completed marker file output oxo-flow targets files, not directories
scenario rendered at run time from config/scenario.yaml (yte template) pre-rendered resources/scenarios/SRR702070_group.yaml for the default sample group; the template is kept verbatim at config/scenario.yaml one scenario (purity 1.0) in the default path
download_vep_plugins.py with a hard-coded Ensembl variation FTP list and fallback the --release/--output/--log argv variant of the same wrapper port one release (111), one output dir; the FTP fallback list was dropped as dead code in the default path
wrapper-utils based rules (calls, tables, report) plain python scripts/*.py argv ports of the same wrappers wrapper-utils is a Snakemake runtime; the ported scripts keep the wrapper logic verbatim
filter_odds not ported not reachable in the default path (filter: present only); the population/burden branches consume gather_annotated_calls instead (ported in population.oxoflow)
template oncoprint views (gene_oncoprint / variant_oncoprints datasets) empty (upstream defaults with a single group) prepare_oncoprint itself runs and feeds the label-sorting table, exactly like upstream
vembrane filter/table expressions evaluated from Python at run time precomputed literal expression/header (34 columns) same semantics, evaluated once
upstream config/units.tsv absolute /projects/... read paths config.reads_dir + sample group fixture paths portability
Snakemake temp() outputs temporary = true engine equivalent
per-rule conda environments one env per tool pin set (envs/) same packages, same pins, consolidated
snakemake before_update/update flags (population db) no input edge; the db path is read/written as-is oxo-flow has no such flags; a declared input would create a DAG cycle (validate rejects it)
snakemake temp() outputs of the gated branch modules plain outputs (temporary = true where the default path used it) see the module headers; join_mutational_signatures writes with > instead of the upstream >> because the engine does not pre-delete outputs
snakemake script API (snakemake.input/output/params) in the 6 branch scripts argv ports (--output/--log flags, comma-joined lists) same logic verbatim, cf. the default-path script ports
chm sample group vertical slice (benchmarking) not ported the ported CHM-eval flow (chm_eval_sample ... chm_eval) re-derives the CHM1 FASTQs, but the chm sample is not in the port's config/samples.tsv, so the chm reads do not flow through mapping -> calling -> control_fdr; rename_chromosomes/chm_eval keep orphan inputs (validate warns, like upstream without the chm sample)
consensus-read calling (calc_consensus_reads flow) consensus.oxoflow, gated on consensus_activate upstream switches the recalibrate_base_qualities/apply_bqsr input via get_recalibrate_quality_input; the port models this as gated duplicate rules with the same outputs and exclusive when gates (!consensus_activate vs consensus_activate)
annotate_dgidb annotation::annotate_dgidb, gated on dgidb_activate + annotation_selection upstream get_final_selected_annotation switches the annotated callset consumed by filtering and the final-calls chain; the port exposes the same selection as config.annotation_selection
filter_offtarget_variants (wrapper v2.3.2/bio/bcftools/filter, params.extra="") pass-through bcftools filter -o/-O b on the fixed calls; the regions/index inputs are declared (as upstream) so get_target_regions and the candidate indexes exist pre-scatter the pinned wrapper consumes only input[0] (verified against its source); the actual target-region restriction is the filter_group_regions bedtools intersect below
target_regions list config single BED path (config.target_regions) upstream merges one or more files; the port freezes one path
filter_group_regions get_filter_targets (bedtools intersect) same command inline in the two filter_group_regions_* rules byte-identical output; intersect branch only when target_regions is set
per-group calling column of config/samples.tsv [sample_groups.metadata] calling on each group fusions continuation rules gate on wildcard.calling == "fusions" || "variants,fusions"
get_candidate_calls for caller=arriba (UNFILTERED group concat) + get_varlociraptor_params (propagate-info-fields extra) calling::varlociraptor_preprocess_arriba/varlociraptor_call_arriba consuming results/candidate-calls/arriba/{group}/{group}.bcf command text identical; the arriba path has no scatter fan-out (no scatteritem)
scatter_candidates/filter_group_regions conditional inputs (Python if config.get("target_regions", None)) optional = "any" input pairs + if [ -n "{config.target_regions}" ] shell switch engine equivalent of the upstream input selection
upstream get_target_regions chr-strip (awk '{sub("^chr","",$0); print}') verbatim target BEDs must be chr-less (Ensembl GRCh38 primary assembly); chr-prefixed files fail closed, exactly as upstream

Created on 2026-08-15 — this port may lag behind upstream releases. See the repository's NOTICE for full attribution.