Next, we selected significant PRS and frequency band pairs (low theta, high theta, low alpha, high alpha, low beta, mid beta, high beta) for subsequent investigation. This selection was performed separately among males and females. For each frequency band, and PRS, results from the 96 associations of the PRS and EEG coherence pairs were divided into three age groups (12–18, 18–25, 25–31). For each age group, the first-quartile p-value was calculated. PRS and frequency bands were selected when at least eight EEG coherence pairs with p-values < 10−6 were present in at least one of the three age ranges. The use of the first quartile p-value is a conservative method to avoid isolated false positives and to ensure biological meaningfulness in a procedure which produces large numbers of correlated results because of the use of overlapping data sets, and the relatively high degree of phenotypic correlation among EEG coherences. Among the frequency bands that met this stringent criteria, the entire association time series of each EEG coherence pair and PRS was examined, and a false discovery rate estimation was applied. All analyses were conducted in MATLAB.
