The Segmentome: a novel representation of the DNA methylome identified by a Bayesian HMM framework

An unsupervised Bayesian HMM infers a segment-level “segmentome” from WGBS, partitioning human genomes into unmethylated, methylated, and transition zones, enabling cell-type grouping.

• Why it matters: Most methylome work emphasizes differentially methylated regions, leaving within-methylome zone organization and whether segment-level descriptors reflect cell identity unclear.
• What they did: The model uses CpG methylation levels plus read-pattern entropy, local CpG correlation, and CpG count to infer zone boundaries from WGBS, then profiles 207 samples across 39 human cell types.
• The result: Segmentome profiles grouped samples by cell type with adjusted Rand index 0.55 and adjusted mutual information 0.72, and top segmentome biomarkers determined cell-of-origin in cell-free DNA and white blood cell samples.

AI summary of the abstract · may contain errors · not medical advice · read the paper.