Fig. 3.

Comparison of birthweight epigenome-wide association analysis results with or without controlling for mQTL (450 K array). Comparison of two EWAS model’s (mQTL-model and non-mQTL-model) results analyzing the correlation between birthweight and DNA methylation for a total of 129,072 CpGs matched to mQTLs. One model included mQTL for each CpG as a covariate (y-axis), while the other did not (x-axis). For the same CpG, regression effect sizes and P values were compared in these two models. A regression β coefficients from these two models were compared for CpGs on the 450 K array. The 15 CpGs that were significantly associated with birthweight were marked red, and CpGs that were not significant were marked turquoise. Red lines represent X = 0, Y = 0 and X = Y, respectively. Blue line represents regression line between Y and X. B Negative log P values from these two models were compared, with significant CpGs marked red. Similar to A, X = Y is shown with a red line and blue line shows regression line between Y and X