Association of PRS with EEG coherence trajectories were calculated as described in detail in Chorlian et al. [54]. Briefly, to test for association between each PRS and EEG coherence phenotype (27 coherence pairs, 7 frequency bands, 3 PRS levels), a local linear (non-parametric) regression model was calculated, including the PRS as the predictor and the inverse hyperbolic tangent of the coherence measure as the dependent variable. Using an additive model, 96 age centers, 2/10ths of a year apart from ages 12 to 31, were utilized to estimate the age specific association of the PRS with EEG coherence phenotypes, using an additive model. An Epanechnikov kernel with a bandwidth of 0.6 was utilized. Since this method uses sliding age windows, there was considerable overlap in the data used in calculations for nearby ages.
