Fig. 3

Quantile–quantile (Q-Q) plots for post-meta-analysis models for association between seasons of birth and DNA methylation (primary analyses) for at-birth (A and B) and childhood (C and D) samples. The Q–Q plots were generated by plotting observed p values (y-axis) against the expected uniform distribution of p values under the null hypothesis of no association (x-axis). Lambda and bias were estimated using BACON method [37]. All cohort-specific EWAS analyses were adjusted for gender of the child, gestational age at delivery, maternal age at delivery, maternal smoking during pregnancy, maternal socio-economic status, batch, child’s age at the time of sample collection and estimated cell proportions