Circulating cell-free DNA (cfDNA) encodes a wealth of biological information, reflecting both tissue-of-origin signals and the underlying cellular states from which it is released. Advances in genome-wide profiling have enabled the deep interrogation of this hidden information, catalysing rapid progress in liquid biopsy for cancer detection and monitoring. Early approaches focused on detecting somatic mutations but were constrained by the limited number of recurrent mutations and the low tumour fraction (that is, a low proportion of tumour-derived DNA within total cfDNA). These limitations prompted a pivot towards alternative signal types, particularly epigenetic features such as DNA methylation. Over the past decade, the field of liquid biopsy research has evolved from relying on sparse mutational signals to harnessing the rich epigenetic landscapes that now underpin next-generation approaches.

A foundational advance came from the recognition that DNA methylation is not simply a collection of independent CpG sites but is instead organized into coordinated, higher-order structures across adjacent genomic regions. In 2017, the systematic analysis of whole-genome bisulfite sequencing datasets by Guo et al. identified nearly 150,000 methylation haplotype blocks (MHBs). These regions of tightly coupled CpG methylation across individual DNA molecules exhibit highly structured, non-random methylation patterns. Importantly, this study introduced the concept of methylation haplotype load (MHL), a quantitative metric that captures coordinated methylation patterns across genomic regions using single-read information. This shift from single-CpG site to block-level analysis can increase robustness and thus represented a key conceptual advance in methylome analysis.