Summary
A bioRxiv preprint from Queen Mary University of London presents a genome-wide framework that uses native Oxford Nanopore reads to measure molecule-level DNA methylation variation and co-methylation. In human lymphoblastoid cell data, the analysis found patterns hidden by site-level averages.
A preprint from researchers at Queen Mary University of London introduces a genome-wide method for studying DNA methylation at the level of individual DNA molecules. Using native Oxford Nanopore Technology (ONT) sequencing, the framework measures variation between molecules and identifies methylation sites and regions whose states occur together on the same molecules.
The authors tested the approach with data from human lymphoblastoid cell (LCL) cells. They report that substantial methylation heterogeneity was hidden when the data were reduced to averages across sequencing reads. The framework also revealed shared and sex-specific co-methylation patterns and was integrated into the DMRcaller R/Bioconductor package. The work is presented as a bioRxiv preprint.
From average methylation to individual molecules
DNA methylation is an epigenetic modification involved in gene regulation. Many analyses summarise methylation at a genomic site by averaging observations from multiple DNA molecules. That summary is useful, but different underlying molecule-level patterns can produce a similar average.
For example, an average methylation level could arise because nearly every molecule carries an intermediate state, or because two groups of molecules carry distinctly different states. A site-level summary can make those cases look alike. Long native DNA molecules sequenced with ONT retain methylation information across multiple positions on the same molecule, allowing the relationships between those positions to be analysed directly.
The new framework focuses on two types of information that site-level summaries can miss. It detects Variable Methylated Domains (VMDs) and Variable Methylated Regions (VMRs) to quantify heterogeneity between individual molecules. It also identifies Co-Methylated Positions (CMPs) and Co-Methylated Regions (CMRs), where methylation states are coordinated across positions or regions observed on the same DNA molecules.
This distinction matters because methylation at nearby sites may not be independent. A pattern that appears only when several positions are considered together can provide information about how regulatory regions are organised, rather than simply how methylated each site is on average.
Findings from the human cell-data analysis
In the human LCL-cell data, the researchers found molecule-level methylation variation that was masked by conventional site-level summaries. Their co-methylation analysis identified coordinated patterns between CpG sites and genomic regions, including patterns shared across the analysed data and patterns associated with sex.
The study also used CMPs to detect pairs of transcription factors predicted to have coordinated binding. In this analysis, the coordination is inferred from the methylation states found together on individual molecules; the abstract describes the transcription-factor relationships as predictions rather than as direct binding measurements.
The framework is designed for scalable, genome-wide analysis and has been incorporated into DMRcaller, an R/Bioconductor package. That integration could make the method easier to apply within existing computational workflows for methylation and differential-methylation analysis.
The evidence currently comes from a computational framework demonstrated with human LCL-cell data. The source does not report a sample count or quantitative performance measures in its abstract, and it does not describe clinical outcomes or validation across other tissues. The biological significance of the reported sex-specific patterns and predicted transcription-factor coordination therefore remains tied to follow-up analyses beyond this initial preprint demonstration.