Estimating DNA methylation heterogeneity in bisulfite sequencing data. Homo sapiens

We extended the mathematical models of measuring biodiversity to estimate DNA methylation heterogeneity in a cell population. We propose a model-based approach (abundance-based, phylogeny-based and pairwise similarity-based heterogeneity) and consider similarity in DNA methylation patterns from individual cells to evaluate heterogeneity that overcomes biases due to missing data. We also applied commonly used non-model based method (methylation entropy) and other reported methods of estimating methylation heterogeneity such as single-cell based approach to evaluate methylation heterogeineity. Overall design: The performance of model-based methods for estimation of methylation heterogeneity is exemplified by artificial data, and compared by real datasets from normal and tumor tissue profiled by reduced representation bisulfite sequencing (RRBS).