Decoding coupled epigenetic-fragmentomic signals in circulating cell-free DNA for interpretable cancer diagnosis.
4 affiliations
- 1College of Informatics, Huazhong Agricultural University, Wuhan, 430070, People's Republic of China.
- 2Medical Research Center, the Eighth Affiliated Hospital, Sun Yat-Sen University, Shenzhen, 518033, People's Republic of China. [email protected].
- 3College of Informatics, Huazhong Agricultural University, Wuhan, 430070, People's Republic of China. [email protected].
- 4College of Informatics, Huazhong Agricultural University, Wuhan, 430070, People's Republic of China. [email protected].
Abstract
Background
Plasma cell-free DNA (cfDNA) whole-genome methylation sequencing enables simultaneous characterization of DNA methylation and fragmentomic features for cancer detection. However, it remains unclear whether methylation library preparation faithfully preserves fragmentomic signals, how methylation and fragmentation are biologically coupled, and how these two modalities can be systematically integrated.Methods
We systematically evaluated the preservation of cfDNA fragmentomic features in whole-genome bisulfite sequencing (WGBS) and enzymatic methyl sequencing (EM-seq) relative to conventional whole-genome sequencing (WGS). We next investigated the biological relationship between methylation and fragmentation and delineated methylation haplotype blocks (MHBs) directly from plasma cfDNA as biologically interpretable units for organizing multimodal cfDNA signals. Guided by these findings, we developed multimodal diagnostic models integrating complementary methylation and fragmentomic features.Results
Canonical cfDNA fragmentomic patterns were partially preserved across methylation sequencing protocols, although the degree of preservation varied among individual features. DNA methylation levels were associated with characteristic fragment length and end-motif patterns. Plasma-derived MHBs exhibited tissue-associated regulatory features and distinct patterns of chromatin accessibility, repeat enrichment, and methylation heterogeneity, providing biologically interpretable units for integrated cfDNA analysis. Within these regions, methylation and fragmentomic signals exhibited coordinated yet complementary patterns. Leveraging this framework, we developed Fragmentomics And Methylation Ensemble (FAME), an interpretable multimodal model integrating MHB-centered methylation and fragmentomic features for cancer detection and tissue-of-origin prediction, and further extended it with complementary genome-wide methylation and fragmentomic features to generate a genome-wide-augmented FAME model (FAME-GW). Across cross-validation, independent validation, and comparisons with THEMIS (THorough Epigenetic Marker Integration Solution), SPOT-MAS (screening for the presence of tumor by methylation and size), and DELFI (DNA evaluation of fragments for early interception), FAME-based models achieved robust diagnostic performance, with FAME providing a compact and interpretable representation and FAME-GW delivering the strongest overall predictive performance.Conclusions
Our study demonstrates that cfDNA whole-genome methylation sequencing partially preserves informative fragmentomic signals and establishes plasma-derived MHBs as biologically interpretable units for organizing and integrating complementary methylation and fragmentomic information. This framework provides insight into the coupling of epigenetic and fragmentomic information and an effective strategy for interpretable cfDNA-based cancer diagnosis.
Related
- Decoding chromatin-associated RNA (caRNA) interaction networks: from genome-wide dynamic capture to epigenetic mechanisms in plants and animals.
- SHERLOC: An interpretable deep learning model for longitudinal circulating tumor DNA data in survival analysis
- Circulating epigenetic and epitranscriptomic biomarkers identify EGFR status, predict TKI response and reveal therapeutic vulnerabilities in NSCLC
- Interpretable attention-deficit hyperactivity disorder diagnosis from functional magnetic resonance imaging via hybrid convolutional neural network–transformer learning and multi-level graph-based comprehensibility
- Circulating biomarkers in the diagnosis and management of neuroendocrine neoplasms
BibTeX
@article{wang2026decoding,
title = {{Decoding coupled epigenetic-fragmentomic signals in circulating cell-free DNA for interpretable cancer diagnosis.}},
author = {Yunze Wang and Yuying Hou and Xinyu Liu and Qingjiao Li and Wen Zhang and Xionghui Zhou},
journal = {Genome medicine},
year = {2026},
month = oct,
doi = {10.1186/s13073-026-01793-w},
url = {https://europepmc.org/article/MED/42888434},
note = {CC BY-NC-ND},
}
RIS
TY - JOUR
TI - Decoding coupled epigenetic-fragmentomic signals in circulating cell-free DNA for interpretable cancer diagnosis.
AU - Yunze Wang
AU - Yuying Hou
AU - Xinyu Liu
AU - Qingjiao Li
AU - Wen Zhang
AU - Xionghui Zhou
PY - 2026
DA - 2026/10/15
JO - Genome medicine
DO - 10.1186/s13073-026-01793-w
UR - https://europepmc.org/article/MED/42888434
AB - Background Plasma cell-free DNA (cfDNA) whole-genome methylation sequencing enables simultaneous characterization of DNA methylation and fragmentomic features for cancer detection. However, it remains unclear whether methylation library preparation faithfully preserves fragmentomic signals, how methylation and fragmentation are biologically coupled, and how these two modalities can be systematically integrated.Methods We systematically evaluated the preservation of cfDNA fragmentomic features in whole-genome bisulfite sequencing (WGBS) and enzymatic methyl sequencing (EM-seq) relative to conventional whole-genome sequencing (WGS). We next investigated the biological relationship between methylation and fragmentation and delineated methylation haplotype blocks (MHBs) directly from plasma cfDNA as biologically interpretable units for organizing multimodal cfDNA signals. Guided by these findings, we developed multimodal diagnostic models integrating complementary methylation and fragmentomic features.Results Canonical cfDNA fragmentomic patterns were partially preserved across methylation sequencing protocols, although the degree of preservation varied among individual features. DNA methylation levels were associated with characteristic fragment length and end-motif patterns. Plasma-derived MHBs exhibited tissue-associated regulatory features and distinct patterns of chromatin accessibility, repeat enrichment, and methylation heterogeneity, providing biologically interpretable units for integrated cfDNA analysis. Within these regions, methylation and fragmentomic signals exhibited coordinated yet complementary patterns. Leveraging this framework, we developed Fragmentomics And Methylation Ensemble (FAME), an interpretable multimodal model integrating MHB-centered methylation and fragmentomic features for cancer detection and tissue-of-origin prediction, and further extended it with complementary genome-wide methylation and fragmentomic features to generate a genome-wide-augmented FAME model (FAME-GW). Across cross-validation, independent validation, and comparisons with THEMIS (THorough Epigenetic Marker Integration Solution), SPOT-MAS (screening for the presence of tumor by methylation and size), and DELFI (DNA evaluation of fragments for early interception), FAME-based models achieved robust diagnostic performance, with FAME providing a compact and interpretable representation and FAME-GW delivering the strongest overall predictive performance.Conclusions Our study demonstrates that cfDNA whole-genome methylation sequencing partially preserves informative fragmentomic signals and establishes plasma-derived MHBs as biologically interpretable units for organizing and integrating complementary methylation and fragmentomic information. This framework provides insight into the coupling of epigenetic and fragmentomic information and an effective strategy for interpretable cfDNA-based cancer diagnosis.
SN - 1756-994X
ER -