Summary
A bioRxiv preprint maps nucleosome patterns across 410 tumours from 23 cancer types and reports that tissue-derived patterns could distinguish cancers using circulating cell-free DNA. The findings point to a possible source of information for non-invasive cancer detection, but the reported results are from research analyses rather than a demonstrated clinical test.
Researchers have mapped nucleosome patterns across 410 primary tumours representing 23 cancer types and report that some of those patterns can also be detected in circulating cell-free DNA (cfDNA). The bioRxiv preprint, posted on October 18, 2026, reports AUROC values of 0.877 for classification among three cancer types and 0.930 for colorectal cancer detection; the latter was 0.773 in an independent validation cohort.
The study uses nucleosome organization as an additional source of information about tumours. Its findings suggest that DNA patterns in blood could help distinguish cancer types, although the reported results do not establish a ready-to-use clinical test.
Mapping nucleosome patterns across cancers
DNA in cells is packaged around proteins into structures called nucleosomes. Their positions and occupancy—the extent to which DNA is wrapped around them—are features of chromatin, the DNA-and-protein material that makes up chromosomes. These patterns can vary with how genes and other regulatory regions are organized.
The researchers analysed ATAC-seq data from 410 primary tumours spanning 23 cancer types. ATAC-seq measures how accessible regions of chromatin are. The study also checked its atlas against independent MNase-seq data, a different method for examining nucleosome organization.
The authors report both patterns shared across cancers and patterns enriched in particular cancer types. Nucleosome occupancy offered information complementary to chromatin accessibility for tumour stratification and revealed molecular subgroups within cancers, including subgroups associated with Basal-like breast cancer. In hepatocellular carcinoma, the analysis identified cancer-associated chromatin changes linked to regulatory pathways, transcription-factor motifs and candidate genes.
Signals in cell-free DNA
CfDNA consists of DNA fragments circulating in blood. Because the study found tissue-derived nucleosome signatures in cfDNA, the authors tested whether those signatures could help classify cancer types. They report an AUROC of 0.877 for classification among three cancer types and an AUROC of 0.930 for colorectal cancer detection, with an AUROC of 0.773 in an independent validation cohort.
AUROC measures how well a model separates groups across possible decision thresholds; it is not the same as the accuracy of a clinical test in a particular population. The abstract does not give the cfDNA cohort sizes or describe the clinical study design, so the reported figures should be understood as results from the analysed datasets, not as evidence of population-screening performance.
The team also describes OncoNucAtlas, a publicly accessible database for exploring the nucleosome landscapes. Together, the atlas and cfDNA analyses provide a resource for studying tumour heterogeneity and for further evaluation of nucleosome-based signals in non-invasive cancer detection.