Prior to quantitative trait loci analysis each trait was adjusted using the available biological and methodological covariates in an attempt to remove the influence of these potentially confounding affects. In R each trait was regressed using the following model:Where Y is the trait profile (log2 normalized mRNA expression intensities and raw values of CpG DNA methylation) and X1 … Xn represent the biological covariates Age and Gender and the methodological covariates post mortem interval (PMI), which Brain Bank the samples was from and which preparation/hybridization batch the sample was processed in. Within this model gender, tissue bank and batch where treated as categorical covariates. After fitting each trait to the model the residuals from the model are kept and represent the trait in following analyses. Thus variance attributable to gender, age, post-mortem interval, tissue source and hybridization batch are removed prior to QTL analysis. These covariate data are available in Table S1. Histograms showing the proportion of traits that are potentially impacted by these covariates are shown in Figure S4 (CpG) and Figure S5 (mRNA).
