Nextflow Outputs

Output files produced by the Krewlyzer Nextflow pipeline.

WGS Output Directory

results/
├── {sample}.bed.gz                         # Extracted fragments
├── {sample}.bed.gz.tbi                     # Tabix index
├── {sample}.metadata.tsv                   # Run metadata and QC metrics (tabular)
├── {sample}.correction_factors.tsv         # GC correction factors
├── {sample}.EndMotif.tsv                   # 4-mer end motif frequencies
├── {sample}.EndMotif1mer.tsv               # 1-mer end motif + Jagged Index
├── {sample}.BreakPointMotif.tsv            # Breakpoint motif frequencies
├── {sample}.MDS.tsv                        # Motif Diversity Score
├── {sample}.FSC.tsv                        # Fragment size coverage (off-target)
├── {sample}.FSR.tsv                        # Fragment size ratio
├── {sample}.FSD.tsv                        # Fragment size distribution (per arm)
├── {sample}.WPS.parquet                    # WPS nucleosome profiles (foreground)
├── {sample}.WPS_background.parquet         # WPS Alu stacking (background)
├── {sample}.OCF.tsv                        # OCF tissue-of-origin scores
├── {sample}.OCF.sync.tsv                   # OCF sync scores (detail)
├── {sample}.TFBS.tsv                       # TFBS entropy (808 TFs)
├── {sample}.ATAC.tsv                       # ATAC entropy (23 cancer types)
└── {sample}.features.json                  # Unified ML features (--generate_json)

Panel Mode Outputs (with --assay or --targets)

When assay or targets are provided, additional on-target and panel-specific files are generated:

results/
├── ... (all WGS outputs above) ...
│
│── # GC Correction
├── {sample}.correction_factors.ontarget.tsv    # On-target GC factors
│
│── # Motif (on-target split)
├── {sample}.EndMotif.ontarget.tsv
├── {sample}.BreakPointMotif.ontarget.tsv
├── {sample}.MDS.ontarget.tsv
│
│── # FSC (gene-centric + regions)
├── {sample}.FSC.ontarget.tsv                   # On-target FSC
├── {sample}.FSC.gene.tsv                       # Gene-level FSC (e.g., 146 genes for xs2)
├── {sample}.FSC.regions.tsv                    # Per-exon/target FSC
├── {sample}.FSC.regions.e1only.tsv             # E1-only FSC (first exon per gene)
│
│── # FSD (on-target split)
├── {sample}.FSD.ontarget.tsv
│
│── # WPS (panel-specific anchors)
├── {sample}.WPS.panel.parquet                  # Panel gene WPS profiles
│
│── # OCF (on/off-target + panel-filtered OCRs)
├── {sample}.OCF.ontarget.tsv
├── {sample}.OCF.ontarget.sync.tsv
├── {sample}.OCF.offtarget.tsv
├── {sample}.OCF.offtarget.sync.tsv
│
│── # TFBS/ATAC (on-target + sync)
├── {sample}.TFBS.ontarget.tsv
├── {sample}.TFBS.sync.tsv
├── {sample}.TFBS.ontarget.sync.tsv
├── {sample}.ATAC.ontarget.tsv
├── {sample}.ATAC.sync.tsv
├── {sample}.ATAC.ontarget.sync.tsv
│
│── # Region MDS (per-gene/exon)
├── {sample}.MDS.exon.tsv                       # Per-exon MDS scores
└── {sample}.MDS.gene.tsv                       # Gene-level aggregated MDS

mFSD Variant Outputs (with VCF/MAF in samplesheet)

results/
├── {sample}.mFSD.tsv                       # Per-variant mFSD summary
└── {sample}.mFSD.distributions.tsv         # Per-variant size distributions (optional)

UXM Methylation Outputs (with meth_bam in samplesheet)

results/
└── {sample}.UXM.tsv                        # Fragment-level methylation

Validation Artifacts (Parquet runs)

Written per sample by run-all, and gathered once per cohort:

File	Scope	Contents
{sample}.validation.json	sample	Contract findings for that sample
{sample}.fingerprint.json	sample	~20 KB summary — a hash and two counts per column
cohort.validation.json	cohort	Cross-sample degeneracy findings

The split exists because degeneracy is inherently cross-sample: every sample can pass on its own while a metric is constant across all of them. A sample directory is ~1.5 GB, so the gather step compares fingerprints rather than re-reading tables.

Skipped entirely for tsv-only runs — see --output_format in Parameters.

Tool-level mode

--use_runall false produces no fingerprints, so cohort validation is skipped rather than run on a partial set, which would report degeneracy findings that are artefacts of the missing samples.

---

Output Changes in 0.9.0

Six output families changed value semantics. Values are not comparable across this boundary, and PON baselines built on uncollapsed input should be rebuilt.

Output	Change
Every positional family — WPS, OCF, TFBS/ATAC, MDS.exon/gene, FSC.gene/regions	Fragment coordinates corrected when R1 is the rightmost mate (~48% of reads on uncollapsed input)
FSC.gene.*, FSC.regions.*	Size bands aligned to the genome-bin bands; new ultra_long and ultra_long_ratio
FSC.regions.*	New strand, is_e1, is_alt_e1 columns
FSC.regions.e1only.*	Selects on the E1 flags; genes with no annotated first exon are omitted rather than represented by an internal exon
MDS.gene.mds_e1, mds_e1_z	Strand-aware on panel data for the first time; NaN where the gene has no E1, instead of a fabricated 0.0
WPS_background.*	NRL family is data-dependent; new nrl_at_band_limit marks a right-censored estimate

---

Available Modules

Module	Description
KREWLYZER_EXTRACT	Fragment extraction
KREWLYZER_FSC	Fragment Size Coverage
KREWLYZER_FSR	Fragment Size Ratio
KREWLYZER_FSD	Fragment Size Distribution
KREWLYZER_WPS	Windowed Protection Score
KREWLYZER_OCF	Orientation cfDNA Fragmentation
KREWLYZER_MOTIF	End Motif & MDS
KREWLYZER_REGION_ENTROPY	TFBS/ATAC entropy
KREWLYZER_REGION_MDS	Per-gene MDS
KREWLYZER_UXM	Methylation
KREWLYZER_MFSD	Mutant Fragment Size
KREWLYZER_RUNALL	Full pipeline
KREWLYZER_BUILD_PON	Build PON