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 study workflow — from plasma collection and cfDNA extraction through low-pass WGS to fragment size, end motif, and nucleosome footprint analysis.

Designing a cfDNA fragmentomics study means making concrete decisions about plasma volume, sequencing depth, control groups, and replicate strategy before the first sample reaches the sequencer. This guide walks through each parameter with the evidence behind the numbers, so you can scope your project accurately and request a quote with confidence.

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

TL;DR

• Plasma input: 1–4 mL plasma per sample; aim for ≥20 ng cfDNA yield. EDTA and Streck tubes are comparable if processed within 8 hours.
• WGS depth: 0.1–1× for copy number and tumor fraction; 1–2× for DELFI-style fragmentation patterns; 5–6× for comprehensive fragmentomics (size, motifs, footprints).
• Controls: Matched buffy coat for germline filtering, plus ≥50 healthy controls for fragmentomics baselines. Spike-in controls for QC.
• Replicates: Minimum n=2–3 biological replicates per group; technical duplicates for reproducibility assessment.
• What to do next: Use the checklist at the end of this guide to assemble your project packet, then request feasibility review.

What Questions Can cfDNA Fragmentomics Answer in Your Research?

cfDNA fragmentomics extracts biological signals from how cell-free DNA breaks apart in the bloodstream — fragment lengths, end sequences, and nucleosome positioning patterns — rather than from individual mutations. This opens analytical possibilities that mutation-based liquid biopsy cannot access.

The biological information encoded in cfDNA fragments

When cells die, nucleases cleave chromatin into cfDNA fragments. The cleavage pattern reflects the chromatin organization of the tissue of origin: nucleosome-protected regions produce longer fragments, linker regions produce shorter ones, and the DNA ends carry signatures of the nucleases responsible (Cristiano et al., 2019). In cancer, these patterns shift measurably — tumor-derived cfDNA tends to be shorter, with altered end motif frequencies and disrupted nucleosome footprints (Ju et al., 2024).

What fragmentomics can reveal that mutation panels cannot

Fragmentomics does not depend on finding a mutation. It detects systemic alterations in chromatin organization that occur even in tumors without druggable mutations. This makes it especially relevant for cancers with low mutation burden, for early detection where ctDNA fractions are below 0.5%, and for treatment monitoring where no trackable mutation has been identified. The DELFI-TF approach has been shown to detect tumor DNA in samples where mutations were undetectable by ddPCR (van 't Erve et al., 2024).

Common study goals and which fragmentomics readouts serve them

Different study goals benefit from different analytical emphases. Cancer screening and early detection studies typically prioritize fragment size ratios and genome-wide fragmentation profiles. Biomarker discovery projects often focus on end motif analysis and nucleosome footprinting at regulatory regions. Treatment monitoring studies leverage the fact that fragmentomics features track tumor fraction without needing a mutation target. For a detailed walk-through of the metrics that appear in a completed report, see our cfDNA fragmentomics report interpretation guide.

How Much Plasma Do You Need? Input Requirements for cfDNA Fragmentomics

Plasma input drives everything downstream — library complexity, achievable depth, and the lower limit of detection for fragmentomics features. Starting with too little plasma is the most common cause of underpowered fragmentomics studies.

Plasma volume and cfDNA yield targets

Standard protocols use 1–4 mL of plasma per extraction. From 1 mL of plasma from healthy individuals, expect roughly 1–10 ng of cfDNA, though cancer patients can yield substantially more due to elevated circulating DNA. A minimum of 5 ng of cfDNA is acceptable for library preparation; 20 ng is preferred and provides more robust libraries (Rickles-Young et al., 2024). For studies where input may be limiting — such as precious archival samples or pediatric collections — discuss low-input protocols with your provider before committing to a study design. The CD Genomics cfDNA fragmentomics service routinely works with input quantities as low as 5 ng.

Collection tube choice and processing window

EDTA and Streck tubes produce comparable fragmentomics results when plasma is separated within 8 hours of collection. Beyond 8 hours, Streck tubes better preserve cfDNA integrity by limiting genomic DNA release from lysed leukocytes. Whichever tube type you choose, consistency across all samples in the study is essential — do not mix tube types within a cohort. Double-centrifugation (900g × 15 min, then 2,500g × 10 min at room temperature) is the standard plasma preparation protocol.

Sample types compatible with fragmentomics

Standard EDTA plasma is the most common input. For studies requiring archival samples, FFPE-derived cfDNA is not suitable for fragmentomics because formalin fixation introduces DNA fragmentation artifacts that obscure endogenous patterns. Fresh or fresh-frozen plasma stored at −80°C is the gold standard. If you are also considering cfRNA or methylation readouts from the same plasma sample, see our companion article on multi-signal liquid biopsy strategies.

What WGS Depth Is Needed for cfDNA Fragmentomics?

Sequencing depth is not one-size-fits-all. The depth you choose determines which fragmentomics features you can reliably measure, and overshooting adds cost without adding decision-relevant information.

Depth tiers and what each one delivers

For most biomarker discovery and treatment monitoring studies, 5–6× coverage provides the best balance of fragmentomics resolution and per-sample cost. At this depth, fragment size distributions, end motif frequencies, and nucleosome footprints at transcription start sites can all be reliably quantified (van 't Erve et al., 2024). Zhang and colleagues demonstrated that even 0.1× coverage can support fragmentomics-based cancer detection when focused on repetitive elements, achieving an AUC of 0.98 (Zhang et al., 2025). However, nucleosome-resolution analyses require higher depth and are more sensitive to coverage fluctuations (Ivanković et al., 2025).

How to choose depth for your specific goal

If your primary goal is to estimate tumor fraction or screen for large-scale copy number alterations across a large cohort, 0.1–1× is sufficient and keeps per-sample costs low. If you need to discover fragmentomics biomarkers — differential end motif usage, regional fragmentation signatures, or nucleosome shifts at specific regulatory elements — aim for 5–6×. If your study will serve as a reference dataset for future assay development, consider 15–30× on a subset of representative samples.

Depth and cost: a directional framework

Per-sample sequencing cost scales roughly linearly with coverage at low depths. A 5× library costs approximately 5–10 times more in sequencing alone than a 0.5× library, though library preparation costs are constant. When scoping a project, confirm with your provider whether the quoted depth is mean genome-wide coverage or target coverage, and whether it includes the reads lost to duplicate removal and quality filtering. The cfDNA fragmentomics WGS service at CD Genomics can accommodate depth requirements from 0.1× to 30× depending on study goals.

Case-Control Design: How to Match Groups and Avoid Confounders

Fragmentomics features are sensitive to biological and technical variables that have nothing to do with cancer. A well-designed case-control framework accounts for these before the data arrives.

Why age matching matters for fragmentomics

cfDNA fragmentation patterns shift with age, independent of disease status. Age-matched controls are therefore essential — a 65-year-old cancer patient should be compared against healthy controls in the same age range, not against 25-year-old donors. This is not a statistical nicety; it directly affects false discovery rates in fragmentomics studies. Where feasible, match controls to cases on age (±5 years), sex, and collection site.

Matched normal DNA: buffy coat or PBMCs

For studies that include mutation or CNV analysis alongside fragmentomics, matched germline DNA from buffy coat or PBMCs is essential for filtering out germline copy number variants and clonal hematopoiesis signals. Sequence the matched normal at comparable depth to the plasma sample. Without a matched normal, it is difficult to distinguish tumor-specific CNVs from benign germline variation, and fragmentomics features associated with hematopoietic clones may be misattributed to the tumor.

Panel of normals for fragmentomics baselines

Fragmentomics features — fragment size ratios, end motif frequencies, and nucleosome coverage patterns — require a healthy reference baseline. A panel of at least 50–55 age-appropriate healthy controls, processed with the same protocol and sequenced at the same depth as the cases, provides this baseline. The panel of normals serves as the denominator for z-score and other statistical comparisons. Larger panels (100+) improve statistical power, especially for rare fragmentomics features.

Longitudinal designs: built-in controls

For treatment monitoring studies, each patient serves as their own control. Pre-treatment, on-treatment, and post-treatment timepoints compared within the same individual eliminate inter-individual variability and substantially increase statistical power. Three or more post-baseline timepoints are recommended for modeling clonal or fragmentomics trajectories.

Replicates: Technical vs Biological — What Each Tells You

Replicates quantify different sources of variation, and conflating them leads to overconfident conclusions — or unnecessary cost.

Figure 2: Technical replicates from the same plasma sample quantify library and sequencing variability; biological replicates from independent subjects capture real population variation in fragmentomics features.

Technical replicates and what they validate

Technical replicates — aliquots of the same plasma sample processed through independent libraries — measure library preparation and sequencing variability. The sWGS validation study by Rickles-Young et al. reported ≥95% agreement across replicate aliquots processed in the same batch (Rickles-Young et al., 2024). For a new fragmentomics study, running technical duplicates on 10–20% of samples provides a robust estimate of technical noise and helps set realistic thresholds for biological signal detection.

Biological replicates and minimum viable power

Biological replicates — independent subjects in the same experimental group — capture real population variation. For discovery cohorts, n=3 per group is a common minimum, but fragmentomics studies often require larger numbers because fragmentation patterns vary substantially between individuals. Power calculations should account for the expected effect size of the fragmentomics feature of interest, which is frequently smaller than the effect sizes seen in mutation-based analyses. If the expected effect is unknown, pilot with n=5–10 per group and use the observed variance to inform full-scale power calculations.

Batch effects and how to design around them

Distribute cases and controls across sequencing batches and lanes — do not run all cases in one batch and all controls in another. Balanced distribution prevents batch effects from being misinterpreted as biological signals. This is especially important for fragmentomics, where subtle shifts in library preparation or sequencing chemistry can alter fragment size profiles and end motif frequencies. Document plate, batch, and lane as covariates in your metadata from the start.

Controls You'll Need: Buffy Coat, Panel of Normals, and Spike-ins

Controls are not an afterthought in fragmentomics — they are the reference frame that makes your data interpretable. Three control types fill three distinct roles.

Matched germline DNA for somatic filtering

Buffy coat or PBMC DNA from the same patient provides the germline genome. Sequence it at the same depth as the plasma cfDNA. Without it, you cannot distinguish somatic CNVs from germline structural variants, and fragmentomics signals from clonal hematopoiesis may be misattributed to the tumor. This is a single-sample cost that prevents systematic errors across the entire dataset.

Panel of normals for fragmentomics baselines

Every fragmentomics metric is interpreted relative to a healthy reference. A panel of normals — 50 or more healthy individuals, age-matched to your study population, processed with identical protocols — provides this baseline. The panel should be sequenced within the same facility using the same library preparation kit and platform. Cross-institution panels require careful batch correction and are best avoided unless the study specifically evaluates cross-site reproducibility.

Spike-in controls and QC standards

Include ERCC or other synthetic spike-in controls at a known concentration in each library. These provide an internal standard for monitoring library preparation efficiency, PCR duplication rates, and cross-sample normalization. When fragmentomics metrics shift unexpectedly across samples, spike-in performance is the first variable to check — it separates biological signal from technical drift.

What Readouts to Expect: From Fragment Size to Nucleosome Footprints

Understanding what a fragmentomics analysis produces helps you specify deliverables clearly and evaluate whether a provider's output matches your study goals. The following readouts represent the core fragmentomics deliverables from low-pass WGS data.

Fragment size distribution and short/long fragment ratio

The fragment size profile — typically a bimodal distribution with peaks at ~167 bp (mononucleosome) and multiples thereof — shifts in cancer toward shorter fragments. The proportion of fragments ≤150 bp is a widely used metric, and the ΔS150 (change in short fragment proportion after end selection) has been shown to be elevated in cancer (Ju et al., 2024). Fragment size ratio plots comparing case vs control distributions are a standard QC and analysis output.

End motif frequencies

The four nucleotides at each cfDNA fragment end are not random — they reflect the cleavage preferences of the nucleases that generated them. CCCA end motifs, associated with DNASE1L3 activity, are the dominant signature in healthy cfDNA. In cancer, CCCA abundance decreases while A/T-rich motifs increase (Ju et al., 2024). End motif profiles are typically reported as 4-mer frequency tables with case-vs-control comparisons for each motif.

Nucleosome footprints and regulatory region coverage

Nucleosome occupancy inferred from cfDNA coverage depth reveals which genomic regions were protected by histones in the cells of origin. Coverage at transcription start sites, enhancers, and transcription factor binding sites differs between healthy and tumor samples, enabling tissue-of-origin inference and regulatory activity profiling. The N-index — the proportion of cfDNA fragments whose ends fall within nucleosome boundaries — is elevated in cancer and forms the basis of the EXCEL diagnostic model (Ju et al., 2024).

CNV and aneuploidy profiles

Read-depth analysis from low-pass WGS detects copy number alterations at arm-level and focal resolution, depending on depth. At 5–6× coverage, CNVs above approximately 5–10 Mb are reliably detectable. Integrating CNV calls with fragmentomics features can improve detection sensitivity beyond either approach alone. For studies where CNV analysis is a primary endpoint, cancer panel sequencing with dedicated CNV backbone coverage may complement fragmentomics WGS.

Integrated fragmentomics scores

Machine learning models trained on multiple fragmentomics features — fragment size, end motifs, nucleosome coverage, and regional fragmentation — produce integrated scores (e.g., DELFI score, EXCEL N-index, DELFI-TF) that combine the diagnostic power of individual features into a single interpretable metric. These scores are typically reported alongside confidence intervals and, where applicable, comparisons to established thresholds.

For a deeper walkthrough of each metric with example plots and interpretation guidance, see our cfDNA fragmentomics report interpretation guide.

Choosing Between Fragmentomics and Targeted Panels for Your Study

Fragmentomics and targeted mutation panels answer different questions, and many studies benefit from both. If your primary goal is to detect known actionable mutations at high sensitivity, a targeted cancer panel is the appropriate tool. If your goal is to detect the presence of cancer through genome-wide fragmentation signals — especially when no trackable mutation is known or when tissue-of-origin information matters — fragmentomics provides complementary information that panel sequencing cannot.

Recent work has shown that fragmentomics features can even be extracted from targeted panel sequencing data, with the normalized depth metric achieving an AUROC of 0.943 across cancer types (Helzer et al., 2025). This opens the possibility of extracting fragmentomics insights from existing panel data, but dedicated low-pass WGS remains the method of choice for comprehensive fragmentomics analysis. For a detailed comparison of these two approaches, see our companion article: cfDNA fragmentomics vs targeted cancer panels.

A Pre-Inquiry Study Design Checklist

Use this checklist to assemble a project packet before contacting a provider. Clear answers to these items reduce back-and-forth and produce more accurate scoping.

1. Study objective: What decision will this study inform? (e.g., biomarker discovery, treatment monitoring, early detection feasibility)
2. Plasma samples: How many samples, what volume per sample, what collection tube type, and when were they collected relative to treatment?
3. Case-control structure: Number of cases and controls, age matching status, availability of matched buffy coat or PBMCs for each case.
4. Sequencing depth target: Which depth tier fits your goal? (0.1–1×, 1–2×, 5–6×, or 15–30×)
5. Replicate plan: How many biological replicates per group? Will you include technical duplicates for a subset?
6. Control samples: Do you have healthy control plasma for a panel of normals? How many?
7. Analysis deliverables: Which readouts are required? Fragment size profiles, end motif analysis, nucleosome footprints, CNV calls, integrated scores?
8. Metadata: Is sample metadata (age, sex, collection site, treatment status, batch) organized in a structured template?
9. Pilot plan: Will you run a pilot with 5–10 representative samples before scaling to the full cohort?

What to do if any item is uncertain

If you are unsure about depth requirements because the expected ctDNA fraction is unknown, or if sample quantity is limited and you need guidance on feasibility, run a small pilot. Select 5–10 representative samples spanning your expected range (e.g., pre-treatment and post-treatment, or early and late stage), and use the pilot data to lock parameters before scaling. The CD Genomics Liquid Biopsy Solutions team can provide feasibility guidance based on your sample types and study goals.

Figure 3: A decision flowchart summarizing the key study design decisions for cfDNA fragmentomics — from sample QC through depth selection to pilot validation and full-scale execution.

FAQ

Can I use archived plasma samples for cfDNA fragmentomics?

Yes, provided the plasma was processed within the appropriate window (within 8 hours for EDTA tubes) and stored at −80°C. Freeze-thaw cycles degrade cfDNA and introduce fragmentation artifacts, so samples with multiple freeze-thaw cycles should be flagged and may require QC before inclusion. FFPE-derived cfDNA is not suitable for fragmentomics due to formalin-induced fragmentation artifacts.

What if my available plasma volume is below the recommended range?

Low-input protocols can work with as little as 0.5 mL of plasma if the cfDNA yield meets the 5 ng minimum for library preparation. However, lower input reduces library complexity and may limit the fragmentomics features that can be reliably quantified. Discuss low-input feasibility with your provider before committing to a study design, and consider running a pilot sample to confirm that QC metrics and fragmentomics signals meet your study's requirements.

How long does a typical cfDNA fragmentomics study take from sample submission to data delivery?

Timelines vary by project scope and provider workload, but a typical study with 50–100 samples at 5–6× coverage can be completed in 6–10 weeks from sample receipt to final data delivery, inclusive of library preparation, sequencing, and bioinformatics analysis. Larger cohorts and deeper sequencing extend timelines proportionally. Confirm turnaround estimates with your provider during scoping.

Is fragmentomics suitable for all cancer types?

Fragmentomics has demonstrated detection capability across a broad range of solid tumor types, including breast, colorectal, lung, ovarian, pancreatic, gastric, liver, and bile duct cancers. Sensitivity varies by cancer type, stage, and ctDNA shedding rate. Cancers with inherently low ctDNA shedding — such as some gliomas and early-stage prostate cancers — present greater detection challenges. The genome-wide nature of fragmentomics makes it broadly applicable, but per-cancer performance should be evaluated in the context of published evidence for the specific tumor type.

Do I need matched normal DNA for fragmentomics analysis?

For fragmentomics-only analysis (fragment size, end motifs, nucleosome footprints), matched normal DNA is not strictly required because these features are analyzed relative to a panel of normals, not individual germline genomes. However, if your study also includes CNV or mutation analysis — which many fragmentomics studies do — matched germline DNA from buffy coat or PBMCs is essential for filtering germline variants. If in doubt, collecting matched normal samples at the time of blood draw is a low-cost insurance policy.

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