Fig. 4

Solo-WCGWs are confounded by cell type heterogeneity. a Using WGBS data encompassing 6 blood cell subtypes from each of 3 donors, we display the density distributions of absolute differences in DNAm between corresponding pairs of blood cell-subtypes (first 7 panels), as indicated. In gray, we depict the distribution for all CpGs with at least 5 reads mapped in at least one of the donors, whereas in red, we depict the corresponding distribution for the subset of solo-WCGWs. In each panel, we give the fraction of CpGs with absolute DNAm difference larger than 0.2 (vertical dashed line), the threshold used to declare a differentially methylated cytosine (DMC). We also provide the corresponding odds ratio (OR) and associated one-tailed P value from a Fisher test. The last panel summarizes the result for all 15 pairs combined. b Unsupervised clustering of 37 fetal tissue samples representing 10 tissue types, over the 6214 HM450k PMD solo-WCGW CpGs (left panel) and over the 163 epiTOC2 CpGs (right panel). c Distribution of the range in DNAm exhibited by the 6214 HM450k PMD solo-WCGWs (left panel) and by the 163 epiTOC2 CpGs, as assessed over the 37 fetal tissue samples from 10 different tissue types