Fig. 2

Logistic regression analysis reveals DMLs that may be predictive of insulin sensitivity status and whose methylation states in monocytes of individuals with IR strongly associate with that of HSCs. a–d Four CpGs were identified as significantly associated with IR outcome using simple logistic regression analysis of the DMLs, clinical data, and immunological data each of which independently predicted outcome. Linear regression analysis was applied to display this relationship between HOMA-IR scores and DNA methylation at each of the four CpGs and their associated genes for a ESRP1 (cg27655935), b an intergenic region 5kb from miRNA596 (cg02000426), c SVOPL (cg10184328), and d SVOPL (cg23085143). Blue dots represent data from IS individuals, and red dots represent data from IR individuals. Red dotted line shows the cut-off value for HOMA-IR. Spearmen correlation coefficient (r) is shown with significance at P < 0.05. e–h. Box-plots shows the monocyte DNA methylation levels of the four independently predictive CpGs in the IS and IR groups along with the methylation states in HSCs for e ESRP1 (cg27655935), f an intergenic region 5kb from miRNA596 (cg02000426), g SVOPL (cg10184328), and h SVOPL (cg23085143). Differences between these groups were determined by Mann-Whitney U test with significance at P < 0.05. Diagrams above each box-plot shows a linear depiction of the associated gene with the position of each CpG indicated by a red triangle. Blue boxes represent exonic sequences with gene orientation marked (5′- ends). ESRP1, Epithelial Splicing Regulator Protein 1; SVOPL, SV2-Related Protein Homolog-Like; HSC, hematopoietic stem cell