Fig. 4

Heat map demonstrating the overlap of shared, top CpG sites across socioeconomic position domains. The CpGs associated (FDR < 0.05) with four socioeconomic position (SEP) domains, composite, education, income, and assets (household), are shown here. We adjusted for false discovery rate (FDR) within eight epigenome-wide association studies using individual CpG-level summary statistics, including only CpGs analyzed across all studies, arriving at 2748 unique CpGs across six studies (FDR < 0.05). Colors indicate the number of associations per CpG per SEP domain, ranging from 0 to 3. For each SEP domain, a CpG received a value of 0 if it did not survive FDR adjustment or was not analyzed in that domain. Individual CpGs were ordered along the y-axis by chromosomal position, though no apparent pattern in chromosomal position was identified. In total, 59 CpGs appeared in two different studies and 5 CpGs appeared in three different studies. For associations shared between more than one study in each column, 36 CpGs associated with education between two studies and 3 CpGs associated with education between three studies. In the income domain, one CpG associated between two different studies. No CpGs were shared between studies for composite and assets. Composite measures associated with the highest number of CpGs (n = 1389), followed by education (n = 1156), income (n = 624), and assets (n = 544). There was little overlap in CpGs between domains, with 88% of CpGs in the income domain having unique signal, followed by 62% for education, 47% for composite, and 21% for assets. See Additional file 1: Table S4 for more details on summary statistics