Disclaimer: CD Genomics provides sequencing and bioinformatics services for research use only, not for clinical or personal diagnosis. This article is for informational purposes only and does not constitute medical advice.

Figure 1: A cfDNA fragmentomics report typically includes fragment size analysis, end motif profiling, nucleosome footprinting, copy number analysis, and one or more integrated scores — each answering a different question about the plasma sample.

You have shipped plasma samples and waited through library preparation, sequencing, and analysis. The report arrives. This guide walks through every section you should expect — what each plot shows, which numbers matter, and how to read the results for study-level decisions without being a bioinformatician.

CD Genomics provides cfDNA fragmentomics (low-pass WGS) services with standardized report deliverables as part of its Liquid Biopsy Solutions platform.

TL;DR

• Fragment size distribution: The histogram with the ~167 bp peak. A left-shift toward shorter fragments (more ≤150 bp) signals tumor-derived cfDNA.
• End motif profiles: 4-mer frequency tables. Reduced CCCA motif abundance indicates altered nuclease activity — a signature of cancer that is independent of mutations.
• Nucleosome footprints: Coverage tracks at transcription start sites and regulatory regions. The N-index quantifies how much fragment ends fall within nucleosome boundaries; lower values indicate more tumor contribution.
• CNV profiles: Genome-wide copy number calls from read depth. ichorCNA-derived tumor fraction estimates are reliable above ~3%.
• Integrated scores: Machine-learning outputs (DELFI score, EXCEL N-index, DELFI-TF) that combine multiple fragmentomics features into a single interpretable number.
• What to do next: Start with the QC section to confirm data quality, then read the integrated score for overall signal, then drill into individual feature sections relevant to your study endpoint.

What a cfDNA Fragmentomics Report Contains: The Sections at a Glance

A standard fragmentomics report from low-pass WGS data contains five to seven analytical sections, plus a QC summary. Each section answers a specific question about the plasma sample. Understanding the structure before drilling into individual plots saves time and prevents misinterpretation.

The typical report structure

Not every study needs all sections. A screening study may focus on the integrated score and fragment size distribution; a biomarker discovery project will use every section. Confirm which sections are included in your scope of work before sample submission. For guidance on study design parameters that determine report content, see our cfDNA fragmentomics study design guide.

QC Summary: The First Section to Read

Before interpreting any biological result, confirm that the data pass QC. A fragmentomics report that skips QC is incomplete — fragmentomics features are sensitive to pre-analytical variables, and a borderline sample can produce misleading signals.

What the QC table should include

These thresholds are consensus values from published validation studies. Rickles-Young et al. reported ≥95% agreement across replicate aliquots processed with standardized sWGS protocols, and the same study demonstrated that adequate cfDNA input and library QC are the strongest predictors of reproducible fragmentomics results (Rickles-Young et al., 2024).

Red flags that warrant a conversation with your provider

A single borderline metric does not necessarily invalidate a sample. But three patterns should trigger discussion: (1) modal fragment size below 140 bp across multiple samples — suggests systematic degradation; (2) mapping rate below 70% — suggests contamination or library failure; (3) cfDNA input below 5 ng combined with duplication rate above 40% — suggests the library is essentially PCR-amplified from too few starting molecules. In all three cases, fragmentomics features may be unreliable and the sample may need to be re-run or excluded.

Fragment Size Distribution: The Core Signal

The fragment size histogram is the most intuitive section of the report — and often the most informative. It shows the length distribution of sequenced cfDNA fragments, typically from ~50 bp to ~550 bp.

How to read the fragment size plot

Healthy plasma cfDNA produces a characteristic pattern: a dominant peak at ~167 bp (DNA wrapped around one nucleosome plus linker), with smaller peaks at ~320 bp (di-nucleosome) and ~480 bp (tri-nucleosome), reflecting the 10-bp periodicity of nucleosome-wound DNA. In cancer, this pattern shifts — the 167 bp peak may be reduced, and the proportion of fragments below 150 bp increases. This left-shift is the most widely used fragmentomics signal for cancer detection (Cristiano et al., 2019).

Short-to-long fragment ratio

The report should include a short-to-long fragment ratio: the number of fragments in the 100-150 bp range divided by the number in the 151-220 bp range. This ratio is typically computed in non-overlapping windows across the genome — DELFI uses 5 Mb windows, producing ~504 data points per sample — and elevated ratios across multiple windows indicate the presence of tumor-derived cfDNA. A genome-wide heatmap showing z-scores of this ratio against a healthy reference panel is a standard DELFI-style output. The CD Genomics cfDNA fragmentomics service includes short-to-long fragment ratio analysis as a core deliverable.

ΔS150: the enriched signal after in silico selection

The EXCEL method introduced by Ju et al. adds a refinement: after measuring the baseline proportion of short fragments (≤150 bp), an in silico end-selection step preferentially retains fragments likely to be tumor-derived, and the same measurement is repeated. The difference — ΔS150 — represents the enrichment of tumor signal after computational selection. A higher ΔS150 indicates stronger tumor-specific fragmentation (Ju et al., 2024).

End Motif Profiles: The Nuclease Signature

Every cfDNA fragment carries a record of the enzyme that cut it, encoded in the few nucleotides at each fragment end. End motif analysis reads these signatures and quantifies which nucleases were active in generating the cfDNA.

What the 4-mer frequency table tells you

The most common end motif analysis reports the frequency of all 256 possible 4-mers (CCCA, CCTG, AAAA, etc.) at fragment ends, normalized to the genomic background. In healthy plasma, CCCA is the dominant motif — it is the cleavage signature of DNASE1L3, the primary nuclease responsible for generating cfDNA in healthy hematopoietic cells. In cancer, CCCA abundance decreases while A/T-rich motifs increase, reflecting reduced DNASE1L3 activity and the contribution of alternative nucleases from tumor cells (Ju et al., 2024).

Motif Diversity Score (MDS)

The MDS is the Shannon entropy of the 256 end-motif frequencies. A low MDS means a few motifs dominate (healthy pattern); a high MDS means many motifs appear at similar frequencies (cancer pattern, reflecting diverse nuclease activity from multiple cell sources). Helzer et al. found that MDS performed particularly well for small cell lung cancer classification, achieving an AUROC of 0.888 from targeted panel data (Helzer et al., 2025).

ΔMCCCA and end selection enrichment

Parallel to ΔS150 for fragment size, the EXCEL model computes ΔMCCCA: the change in CCCA motif abundance after in silico end selection. A decrease in CCCA after selection means tumor-derived fragments carry fewer CCCA ends — the expected pattern. ΔS150 and ΔMCCCA are typically reported together as complementary metrics from the same end-selection pipeline.

Figure 2: Representative report panels — fragment size distribution (left), end motif sequence logo (center), and nucleosome coverage at transcription start sites (right). Each panel captures a distinct dimension of cfDNA biology.

Nucleosome Footprints: Tissue-of-Origin and Regulatory Information

Nucleosome footprinting infers where nucleosomes were positioned in the cells that released the cfDNA. This is the most biologically rich section of the report — and the most sensitive to coverage depth.

Coverage at transcription start sites

A standard footprint analysis plots cfDNA coverage depth around transcription start sites (TSS), typically ±2 kb. In healthy plasma, the nucleosome-depleted region immediately upstream of the TSS shows a characteristic dip in coverage, flanked by well-positioned nucleosomes. In cancer, this pattern is disrupted — the dip may be shallower or shifted — reflecting altered chromatin organization in tumor cells. The pattern is gene-set-specific: housekeeping genes show a different TSS profile from tissue-specific genes, and this difference enables tissue-of-origin inference.

N-index: a single number for nucleosome organization

The N-index — the proportion of cfDNA fragments whose ends fall within computationally defined nucleosome boundaries — is the core quantitative metric from the EXCEL model. In healthy individuals, cfDNA fragment ends cluster at nucleosome boundaries, producing a high N-index. In cancer, fragments are more randomly distributed relative to nucleosome positions, producing a lower N-index. The N-index, combined with ΔS150 and ΔMCCCA, forms the basis of the EXCEL diagnostic classifier (Ju et al., 2024).

Coverage depth requirements for footprinting

Nucleosome footprinting is the most coverage-sensitive fragmentomics analysis. At 0.1-1× depth, only the broadest footprint features are detectable. At 5-6×, TSS coverage profiles and nucleosome positioning become reliably quantifiable. At 15-30×, high-resolution footprinting at individual regulatory elements becomes possible. The report should note the sequencing depth and the resolution limits that apply. For study design decisions about depth, see our cfDNA fragmentomics study design guide.

CNV and Aneuploidy: The Coverage-Based Readout

Read-depth analysis from low-pass WGS detects copy number alterations across the genome. This section of the report bridges fragmentomics and conventional liquid biopsy — it provides the same type of information that a CNV-focused panel would, but across the whole genome.

The genome-wide CNV plot

The standard output is a genome-wide copy number profile: log2 ratio of observed to expected coverage plotted by genomic position, with segmentation calls (via circular binary segmentation or similar) overlaid. Gains and losses are color-coded. At 5-6× coverage, CNVs larger than ~5-10 Mb are reliably detected. Arm-level and chromosome-level aneuploidies are the most robust calls.

Tumor fraction from ichorCNA

ichorCNA is the most commonly used tool for estimating tumor fraction from low-pass WGS data. It models the observed coverage as a mixture of tumor and normal copy number profiles and reports a tumor fraction estimate with a confidence interval. The estimate is reliable above ~3% tumor fraction at 1-6× coverage (Rickles-Young et al., 2024). Below 3%, tumor fraction estimates should be treated as qualitative (detectable vs not detectable) rather than quantitative.

What CNV analysis adds to fragmentomics

CNV calls alone can detect many cancers — but they miss tumors without large-scale copy number changes. Combining CNV with fragment size, end motif, and nucleosome footprint data captures a broader range of tumor biology. In the DELFI framework, CNV information is embedded in the genome-wide fragmentation profile (windows with copy number gain show altered fragment ratios), and the integrated score outperforms either CNV or fragmentation alone (Cristiano et al., 2019).

Integrated Fragmentomics Scores: The Single-Number Summary

Most reports include at least one integrated score — a machine-learning output that combines multiple fragmentomics features into a single numerical result. This is the number to look at first for an overall answer, with individual feature sections providing biological context.

Common integrated scores and what they mean

How to interpret the score for study decisions

An integrated score above the pre-specified threshold (typically set at 95-98% specificity in the reference population) indicates a positive call. Scores near the threshold warrant closer examination of individual feature sections. Scores far above the threshold are high-confidence positives. Scores far below are high-confidence negatives. The report should state the threshold, the specificity at which it was set, and, where possible, the positive predictive value expected in populations with comparable cancer prevalence. For a comparison of fragmentomics with targeted panel approaches — and when integrated scores from both modalities add value — see our companion article on cfDNA fragmentomics vs targeted cancer panels.

How to Use the Report for Study-Level Decisions

A fragmentomics report is not an answer — it is a set of structured results that inform a study decision. The following framework maps common study goals to the most relevant report sections.

By study goal: which sections matter most

What to ask your provider about the report

Before committing to a study, confirm: (1) Which integrated scores are included and at what specificity threshold; (2) Whether the report includes raw data (BAM files, fragment size tables) for re-analysis; (3) Whether longitudinal comparisons are pre-configured or require custom analysis; (4) The reference panel used for z-score calculations and its demographic match to your cohort. Clear answers to these questions before samples are submitted prevent scope mismatches after data delivery. For a deeper treatment of multi-signal approaches that layer fragmentomics with methylation, cfRNA, and other modalities, see our guide on multi-signal liquid biopsy strategies.

Figure 3: An integrated report dashboard combining the core fragmentomics readouts — fragment size distribution, CNV profile, integrated score, and QC summary — into a single-page overview for rapid study-level decision-making.

FAQ

How long does it take to receive a fragmentomics report after sample submission?

A typical timeline is 6-10 weeks from sample receipt to final report delivery for a study with 50-100 samples at 5-6× coverage. This includes cfDNA extraction, library preparation, sequencing, bioinformatics analysis, and report generation. Larger cohorts and deeper sequencing extend timelines proportionally. Confirm turnaround estimates with your provider during project scoping.

What raw data files should I expect alongside the report?

In addition to the interpreted report, you should receive FASTQ files (or aligned BAM files), fragment size tables, 4-mer frequency matrices, CNV segmentation files, and the integrated score outputs with associated metadata. Confirm the specific file formats and delivery method in the scope of work — having raw data enables independent re-analysis and is a marker of a transparent provider.

Can I get fragmentomics metrics from my existing targeted panel data?

Partially. Helzer et al. (2025) demonstrated that fragmentomics features — particularly normalized read depth at captured exons — can be extracted from existing targeted panel sequencing data with an AUROC of 0.943. However, panel-derived fragmentomics provides a subset of the features available from dedicated low-pass WGS, which captures genome-wide fragmentation patterns. If comprehensive fragmentomics analysis is needed, dedicated low-pass WGS is the method of choice.

What if my sample fails QC — can it still be analyzed?

It depends on which QC metric failed and by how much. Samples with cfDNA input slightly below the 5 ng threshold may still produce usable fragment size and end motif data, though nucleosome footprinting resolution will be reduced. Samples with abnormal modal fragment sizes (below 140 bp or above 200 bp) should be reviewed with your provider — the fragmentomics features may reflect pre-analytical artifacts rather than biology. A good provider will flag QC failures before proceeding and discuss options.

How do I compare fragmentomics results across different studies or providers?

Cross-study comparison requires careful attention to protocol differences. The most important variables to align are: sequencing depth, library preparation kit, reference panel for z-score normalization, and the specific fragment size bins used (100-150 vs 151-220 bp is standard but not universal). If you plan to compare results across studies, specify these parameters upfront and ensure both providers use compatible protocols.

Research Use Only. This article is for educational and informational purposes and does not constitute medical or diagnostic advice.