Fragment-level Methylation (UXM)

Command: krewlyzer uxm

Plain English: UXM analyzes methylation patterns to determine which cell types contributed cfDNA. Different tissues have unique methylation signatures—like a "fingerprint" for each cell type.

Use case: Cell-type deconvolution - determine if cfDNA comes from liver, immune cells, tumor, etc.

Purpose

Computes the proportions of Unmethylated (U), Mixed (X), and Methylated (M) fragments per genomic region for cell-type deconvolution.

Processing Flowchart

flowchart LR
    BAM["Bisulfite BAM"] --> RUST["Rust Backend"]
    MARKERS["Methylation Markers"] --> RUST

    RUST --> UXM["UXM.tsv"]

    subgraph "Per Fragment"
        RUST --> CLASS{"CpG Methylation"}
        CLASS -->|"≤25%"| U["Unmethylated (U)"]
        CLASS -->|"25-75%"| X["Mixed (X)"]
        CLASS -->|"≥75%"| M["Methylated (M)"]
    end

Use mouse to pan and zoom

Biological Context

Fragment-level methylation (UXM, Loyfer et al., 2022) reveals cell-of-origin and cancer-specific methylation patterns in cfDNA. Each fragment is classified by its CpG methylation level within marker regions.

Usage

# Basic usage
krewlyzer uxm -i bisulfite.bam -o output_dir/ --genome hg19

# With custom marker file
krewlyzer uxm -i bisulfite.bam -o output/ \
    --mark-input custom_markers.bed.gz

# Paired-end mode
krewlyzer uxm -i bisulfite.bam -o output/ --type PE

CLI Options

Option	Short	Type	Default	Description
`--input`	`-i`	PATH	required	Bisulfite sequencing BAM
`--output`	`-o`	PATH	required	Output directory
`--sample-name`	`-s`	TEXT		Override sample name
`--mark-input`	`-m`	PATH	Auto	Path to genomic marker file
`--type`		TEXT	SE	Sequencing type: SE or PE
`--genome`	`-G`	TEXT	hg19	Genome build (hg19/hg38)
`--verbose`	`-v`	FLAG		Enable verbose logging
`--threads`	`-t`	INT	0	Number of threads (0=all)

Output Format

Output: {sample}.UXM.tsv

Column	Type	Description
`region`	str	Marker region ID
`U_count`	int	Unmethylated fragments
`X_count`	int	Mixed fragments
`M_count`	int	Methylated fragments
`total`	int	Total fragments
`U_frac`	float	U / total
`X_frac`	float	X / total
`M_frac`	float	M / total

Fragment Classification

                   CpG Methylation Level
    ├──────────────┬──────────────┬──────────────┤
    0%            25%            75%           100%
    │      U      │      X       │      M      │
    │ Unmethylated│    Mixed     │ Methylated  │

Class	Threshold	Interpretation
U	≤25% CpGs methylated	Cell-type specific unmethylated
X	25-75%	Heterogeneous/mosaic
M	≥75% CpGs methylated	Stably methylated

Clinical Interpretation

Healthy cfDNA Composition

Based on the Human Methylation Atlas:

Cell Type	Contribution
Megakaryocytes	~31%
Granulocytes	~30%
Monocytes/Macrophages	~20%
Endothelial	~6%
Hepatocytes	~3%
Lymphocytes	~3%

Cancer Detection

Pattern	Interpretation
Altered tissue proportions	Tumor shifts composition
Non-hematopoietic increase	Possible tumor cfDNA
Resolution	Can detect ~0.1% tumor fractions