What's in a name? Context-dependent significance of 'global' methylation measures in human health and disease

The study of DNA methylation in development and disease has 'exploded' as a field in recent years, with three major classes of measurement now routine. These encompass (i) locus-specific, (ii) genome-scale/ wide and (iii) 'global' methylation approaches. Measures of global methylation refer to the level of 5-methylcytosine (5mC) content in a sample relative to total cytosine. Despite this, several other measures are often referred to as 'global', with the underlying assumption that they accurately reflect 5mC content. The two most common surrogate, or proxy, measures include generating a mean or median methylation value from (i) the average measure in thousands of highly repetitive genomic elements and (ii) many thousands to several million primarily unique CpG sites throughout the genome. Numerous lines of evidence suggest the underlying assumption of equivalence of these measures is flawed, with considerable variation in the regulation of different 'flavours' of DNA methylation throughout the genome depending on cell type, differentiation and disease state. As such, the regulation of methylation 'types' is often uncoupled. The emerging picture suggests that no approach can accurately detect all biologically important differences in 5mC variation and distribution in all instances, with this needing to be ascertained on a case-by-case basis. Thus, it is important to clearly elaborate the genomic context and content of DNA methylation being analysed, the sample and developmental stage in which it is being examined and to remember that in most instances, the most common measures are not a true representation of 'global' 5mC content as orginally defined.