The NHGRI GWAS catalog58 was downloaded on accessed on 7 March 2014 from http://www.genome.gov/gwastudies/. After removing duplicate entries, there were 4,887 SNPs with genome-wide significant P values (P < 5 × 10−8), of which 4,546 (93%) were also in the imputed genetic data set of UKBEC after quality control (MAF >5% and imputation R2 >0.50). The disease or trait for each GWAS study was checked and classified as either associated with brain-related phenotypes or not by two of us (M.R. and A.R). Brain-related phenotypes were additionally subclassified according to whether they were an adult-onset neurological disorder or not. For the sentinel marker of each cis-eQTL signal, we calculated the linkage disequilibrium with each of the GWAS SNPs (within a 10-Mb or 1,000-SNP moving window) using the vcftools utility (version 0.1.9) and haplotype data of the European panel (n = 379) from the 1000 Genomes Project (March 2012). If a linkage disequilibrium, r2 ≥ 0.50, was observed, the corresponding GWAS SNP along with the linkage disequilibrium measure was recorded in Supplementary Table 2.
