CFTK Analysis Report

STATS

Sample Statistics

10 samples · 2 groups

Sample		Sequencing & methylation QC				cfDNA QC
Sample	Group	Mapping rate	Duplication rate	Legacy CHH conversion rate	Mean CpG depth	Fragment length peak	β M-score
Control_1	Control	86.893	7.441	99.775	15.04	167 bp	PASS
Control_2	Control	83.268	7.599	99.754	15.39	179 bp	PASS
Control_3	Control	85.921	6.827	99.782	13.83	178 bp	PASS
Control_4	Control	92.441	10.593	99.769	13.53	189 bp	PASS
Control_5	Control	89.924	7.385	99.77	14.56	181 bp	PASS
Disease_1	Disease	89.384	8.509	99.717	16.84	169 bp	PASS
Disease_2	Disease	85.153	7.114	99.609	16.34	169 bp	PASS
Disease_3	Disease	90.189	8.478	99.729	12.46	172 bp	PASS
Disease_4	Disease	83.481	7.698	99.692	17.33	170 bp	PASS
Disease_5	Disease	84.97	7.054	99.697	23.67	165 bp	PASS

PART 01

Data Processing

process

1.1 Trimming

Adapter trimming quality metrics from Trim Galore / Cutadapt.

Filtered Reads⌄

Trimmed Sequence Lengths (3')⌄

ℹ Note: Hover over any line to see the sample name.

1.2 Trimmed QC

FastQC quality metrics on trimmed reads.

Sequence Counts⌄

Sequence Quality Histograms⌄

Per Sequence Quality Scores⌄

Per Sequence GC Content⌄

Per Base N Content⌄

Sequence Length Distribution⌄

Sequence Duplication Levels⌄

Adapter Content⌄

Sequencing Status Checks⌄

1.3 Alignment

Bisulfite sequencing alignment statistics.

Alignment & Deduplication⌄

Note: bwameth does not produce Bismark-style strand alignment or cytosine methylation context plots. Alignment rates are derived from samtools flagstat; deduplication from sambamba markdup.

1.4 M-bias

Per-position methylation bias from MethylDackel mbias. Dataset buttons switch between OT/OB strand and R1/R2. All samples shown simultaneously.

M-bias Plot⌄

1.5 Sequencing QC Summary

Per-sample quality metrics. Click any row to view the recommendation. Column headers are sortable.

Sequencing QC Metrics Table⌄

Filter:Click row for recommendation · Click header to sort

Sample	Status	Mapping rate	Properly paired	Duplication rate	Sequencing error rate	Legacy CHH conversion rate	Global CpG methylation	CpG covered sites	Mean CpG depth
Control_1	PASS	86.9%	100.0%	7.4%	0.789%	99.78%	44.3%	3828321	15.0×
Sample passes all QC thresholds (score 100.0/100). Suitable for downstream analysis.
Control_2	PASS	83.3%	100.0%	7.6%	0.830%	99.75%	52.4%	4262817	15.4×
Sample passes all QC thresholds (score 100.0/100). Suitable for downstream analysis.
Control_3	PASS	85.9%	100.0%	6.8%	0.802%	99.78%	55.3%	2925522	13.8×
Sample passes all QC thresholds (score 100.0/100). Suitable for downstream analysis.
Control_4	PASS	92.4%	100.0%	10.6%	0.786%	99.77%	61.0%	2608848	13.5×
Sample passes all QC thresholds (score 100.0/100). Suitable for downstream analysis.
Control_5	PASS	89.9%	100.0%	7.4%	0.814%	99.77%	56.5%	3621995	14.6×
Sample passes all QC thresholds (score 100.0/100). Suitable for downstream analysis.
Disease_1	PASS	89.4%	100.0%	8.5%	0.733%	99.72%	52.2%	4535257	16.8×
Sample passes all QC thresholds (score 100.0/100). Suitable for downstream analysis.
Disease_2	PASS	85.2%	100.0%	7.1%	0.774%	99.61%	45.5%	4880239	16.3×
Sample passes all QC thresholds (score 100.0/100). Suitable for downstream analysis.
Disease_3	PASS	90.2%	100.0%	8.5%	0.796%	99.73%	59.9%	1389689	12.5×
Sample passes all QC thresholds (score 100.0/100). Suitable for downstream analysis.
Disease_4	PASS	83.5%	100.0%	7.7%	0.851%	99.69%	48.7%	4862884	17.3×
Sample passes all QC thresholds (score 100.0/100). Suitable for downstream analysis.
Disease_5	PASS	85.0%	100.0%	7.1%	0.955%	99.70%	46.4%	5975924	23.7×
Sample passes all QC thresholds (score 100.0/100). Suitable for downstream analysis.

PART 02

cfDNA QC Analysis

qc

2.1 Methylation Distribution

β-value density plot⌄

Sample	β M-score
Control_1	PASS
Control_2	PASS
Control_3	PASS
Control_4	PASS
Control_5	PASS
Disease_1	PASS
Disease_2	PASS
Disease_3	PASS
Disease_4	PASS
Disease_5	PASS

ℹ Note: Human cfDNA shows a bimodal distribution (peaks near 0 and 1). The β M-score assesses whether the CpG methylation distribution is healthily bimodal. β M-score = min(left_peak, right_peak) / mid_density (left_peak: max density at β≤0.15; right_peak: max density at β≥0.85; mid_density: median density at 0.35≤β≤0.65). PASS ≥ 2 — a strong bimodal shape relative to the mid region.

2.2 Fragment Length Distribution

Fragment length plot⌄

Sample	Peak site
Control_1	167 bp
Control_2	179 bp
Control_3	178 bp
Control_4	189 bp
Control_5	181 bp
Disease_1	169 bp
Disease_2	169 bp
Disease_3	172 bp
Disease_4	170 bp
Disease_5	165 bp

Group mean traces are shown by default (x-axis: 50–250 bp). Click a sample in the table to show its fragment-length line on the chart (click again to hide); active samples are highlighted. Hover the chart for exact values.

2.3 Dinucleotide Frequency

Dinucleotide frequency plot⌄

10-bp periodicity of AT- and GC-rich dinucleotides around fragment centres reflects nucleosome positioning.

2.4 PCA

PCA plot⌄

Select modality from dropdown. Hover points for sample names.

Modality:

PART 03

Differential Analysis

diff

3.1 Violin

Violin plots (static)⌄

Modality:

CPG

3.2 Heatmap

DMC heatmaps (static)⌄

Modality:

CPG

3.3 DMR Analysis

DMR Volcano⌄

Each point is one DMR. Hover for gene name and exact values. Dashed line = q-value threshold.

Significant DMRs⌄

PART 04

Fragmentomics

frag

4.1 DELFI

Control

Sample:

Control (mean)

Disease

Sample:

Disease (mean)

Group Comparison

4.2 End Motif

Top 20 4-mer end motifs, shown separately per group. Use each chart's dropdown to switch between the group mean and individual samples.

Control

Sample:

Disease

Sample:

Group comparison: Box plots of the top 20 motifs (by overall mean). Each box summarises the group distribution; individual samples are overlaid as points.

4.3 Cleavage

Control — Group Mean

Disease — Group Mean

Group Comparison

4.4 WPS

Control

Sample:

Control_1

Disease

Sample:

Disease_1

PART 05

MESA Multimodal Modeling

mesa

5.1 Modality Performance

Modality	Clf AUCs (RFC/LR/SVC)	Best Classifier	Best AUC
cpg	[0.9467 0.8267 0.76 0.6133]	RandomForest	0.9467
occupancy	[0.8867 0.5067 0.88 0.5467]	RandomForest	0.8867
wps	[0.7467 0.46 0.8 0.5 ]	SVC	0.8000

5.2 ROC Curve

ROC curves⌄

LOOCV ROC curves per modality. AUC shown in legend. Hover for FPR/TPR values.

5.3 Prediction Heatmap

LOOCV prediction probabilities⌄

Samples sorted by true label then probability. Hover cells for exact values.

5.4 Spearman Correlation

Modality Spearman correlation⌄

Pairwise Spearman ρ of LOOCV prediction scores across modalities.

PART 06

Software & Tools

env

Bioinformatics tools used in this cfDNA methylation analysis pipeline, grouped by processing stage.

Quality Control & Trimming

Tool	Version	Purpose
FastQC	0.12.1	Per-base sequence quality, GC content, adapter, and duplication QC of raw and trimmed reads.
Trim Galore	0.6.10	Adapter and quality trimming of paired-end reads (wraps Cutadapt + FastQC).
Cutadapt	4.9	Adapter sequence removal (invoked by Trim Galore).
MultiQC	1.25	Aggregates FastQC, Trim Galore, and samtools metrics into per-step QC reports.

Alignment & Deduplication

Tool	Version	Purpose
bwa-meth	0.2.7	Bisulfite-aware read alignment to the reference genome.
BWA	0.7.18	Core Burrows-Wheeler alignment engine used by bwa-meth.
SAMtools	1.22.1	BAM sorting, indexing, flagstat, and stats (mapping rate, insert size, error rate).
Sambamba	1.0.1	Duplicate marking and removal on aligned BAM files.

Methylation Extraction

Tool	Version	Purpose
MethylDackel	0.6.1	Per-CpG methylation calling and M-bias estimation. The current default does not extract CHH/CHG contexts.
BEDTools	2.31.1	Fragment extraction (bamtobed), genomic interval operations, and dinucleotide nucleotide content (nuc).

Quality Control Analysis

Tool	Version	Purpose
deepTools	3.5.5	Fragment length distribution via bamPEFragmentSize.

Differential Methylation

Tool	Version	Purpose
metilene	0.2-8	De novo differentially methylated region (DMR) detection.
annotatr	1.32.0	Genomic annotation of DMRs (gene symbol, region type: promoter/intron/exon/CpG island).
R	4.4.1	Statistical environment for DMR annotation and downstream analysis.

Fragmentomics & Modeling

Tool	Version	Purpose
scikit-learn	1.5.2	PCA, classifiers (RandomForest, LogisticRegression, SVC), and LOOCV for MESA multimodal modeling.
NumPy	2.1.2	Numerical computation across all analysis steps.
pandas	2.2.3	Tabular data handling and matrix operations.
SciPy	1.14.1	Statistical tests, KDE density estimation, and Spearman correlation.

Visualization & Reporting

Tool	Version	Purpose
Matplotlib	3.9.2	Static figure generation (methylation distribution, violin, heatmap, ROC, etc.).
Plotly.js	2.35.2	Interactive charts embedded in this HTML report.