Fig. 5

Characterization of differentially methylated CpGs of non-shared environmental origin (eDMR). a eDMRs identified in adipose were associated with genes by overlapping with RefSeq-annotated genes (defined as genic regions including 10 kb downstream of the transcription start site); DeSeq normalized total stranded RNA-seq data were used to determine fold expression changes for RefSeq genes between subcutaneous adipose tissue and T-cells, monocytes, and B-cells. The number of genes significantly (DeSeq-adjusted p < 0.05) overexpressed in adipose versus each hematopoietic cell type (>1-fold, >4-fold, >16-fold higher expression in adipose) was determined for eDMR-associated and all RefSeq genes; the latter to identify dynamic adipose-specific genes. Shown is the ratio of eDMR-associated versus RefSeq genes averaged across the three hematopoietic cell types corrected for background at indicated fold overexpression levels in adipose. Stars indicate z-score-derived p-values <0.001. b Same as (a) but comparing eDMR-associated genes overexpressed in hematopoietic cells to adipose. c (top) UCSC-derived scheme of the HSPA12B RefSeq gene locus and CGIs in the gene region (hg19). (bottom) Zoom into HSPA12B gene region overlapping CGI CpG: 121. For each twin within MZ2, methylation levels are shown for each detected CpG site. The trendl ine was determined using a moving average (period = 2). Methylation values highlighted in black in co-twin 1 indicate significant eDMCs (Fisher’s exact test p < 0.05), the associated eDMR is highlighted in rose. d (top) UCSC-derived scheme of the FAM171A2 RefSeq gene locus and CGIs in the gene region (hg19). (bottom) Zoom into FAM171A2 gene region overlapping CGI CpG: 136. Illustration as in (c)