Fig. 2.

Effect of adjusting ‘slide’ or ‘plate’ as a random effect. True positive rates (TPRs) are consistently higher in a downstream EWAS when variation due to ‘slide’ effects in ARIES (a) and ‘plate’ effects in GOYA (b) are removed using random effects models. Random effects models were applied either probe quantiles along with control variation in FN (‘FN+re’) or during the EWAS (‘FN+ewas.re’). TPRs were estimated by comparison to associations from a large meta-analysis (Joubert et al., 2016)