Features Overview

Krewlyzer provides 11 standalone feature extraction commands plus a unified run-all pipeline.

Quick Comparison

Command	Input	Output	Primary Use Case
`extract`	BAM	`.bed.gz`, `.metadata.json`	Fragment extraction & GC factors
`motif`	BAM	`.EndMotif.tsv`, `.MDS.tsv`	Fragmentation patterns
`fsc`	BED.gz	`.FSC.tsv`	Copy number detection
`fsr`	BED.gz	`.FSR.tsv`	Tumor fraction estimation
`fsd`	BED.gz	`.FSD.tsv`	Size distribution analysis
`wps`	BED.gz	`.WPS.tsv.gz`	Nucleosome positioning
`ocf`	BED.gz	`.OCF.tsv`	Tissue of origin
`region-entropy`	BED.gz	`.TFBS.tsv`, `.ATAC.tsv`	Regulatory region analysis
`uxm`	Bisulfite BAM	`.UXM.tsv`	Methylation deconvolution
`mfsd`	BAM + VCF/MAF	`.mFSD.tsv`	Mutant vs wild-type sizes
`run-all`	BAM	All outputs	Complete analysis

Workflow Diagram

flowchart LR
    BAM[BAM File] --> extract
    extract --> BED[.bed.gz]
    BAM --> motif

    BED --> fsc
    BED --> fsr
    BED --> fsd
    BED --> wps
    BED --> ocf

    BAM --> mfsd
    VCF[VCF/MAF] --> mfsd

    BS_BAM[Bisulfite BAM] --> uxm

    subgraph outputs[Output Files]
        fsc --> FSC[.FSC.tsv]
        fsr --> FSR[.FSR.tsv]
        fsd --> FSD[.FSD.tsv]
        wps --> WPS[.WPS.tsv.gz]
        ocf --> OCF[.OCF.tsv]
        motif --> MOTIF[.EndMotif.tsv]
        mfsd --> MFSD[.mFSD.tsv]
        uxm --> UXM[.UXM.tsv]
    end

    BED --> tfbs[region-entropy]
    tfbs --> TFBS[.TFBS.tsv / .ATAC.tsv]

Use mouse to pan and zoom

Feature Categories

Fragmentation Features

Feature	Biological Signal	Clinical Application
FSC	Fragment coverage by size class	CNV detection, copy number profiling
FSR	Short/Long fragment ratios	Tumor fraction estimation
FSD	Size distribution per arm	Nucleosome patterns, ctDNA detection

Nucleosome & Chromatin

Feature	Biological Signal	Clinical Application
WPS	Nucleosome protection scores	Tissue of origin, gene regulation
OCF	Open chromatin fragmentation	Tissue-specific cfDNA detection
Motif	End motif diversity (MDS)	Fragmentation patterns, cancer detection
Region Entropy	TFBS/ATAC size distribution	Regulatory element alterations, cancer detection

Specialized

Feature	Biological Signal	Clinical Application
mFSD	Mutant vs wild-type sizes	MRD monitoring, ctDNA quantification
UXM	Fragment methylation (U/X/M)	Cell-type deconvolution

Choosing Features

For Cancer Detection

krewlyzer run-all sample.bam -r hg19.fa -o output/
# Focus on: FSR (tumor fraction), FSC (CNV), Motif (MDS)

For Tissue of Origin

# Run WPS and OCF
krewlyzer wps -i sample.bed.gz -o output/
krewlyzer ocf -i sample.bed.gz -o output/

For MRD Monitoring

# Compare mutant vs wild-type fragment sizes
krewlyzer mfsd -i sample.bam -V variants.vcf -o output/

For Methylation Deconvolution

# Requires bisulfite sequencing BAM
krewlyzer uxm bisulfite.bam -o output/

Common Options

All feature commands share these core options:

Option	Description
`-o, --output`	Output directory (required)
`-s, --sample-name`	Override sample name
`-G, --genome`	Genome build: hg19/hg38
`-t, --threads`	Thread count (0=all)
`-v, --verbose`	Enable verbose logging
`-f, --format`	Output format: tsv, parquet, json

See individual feature pages for command-specific options, or JSON Output for format details.

Additional Resources

JSON Output - Unified JSON for ML pipelines (--generate-json)
Panel Mode - MSK-ACCESS and targeted panel analysis (--assay)
PON Building - Creating cohort baselines for z-score normalization