site and 40 kb downstream to the gene's most extreme transcript end site. These boundaries were chosen as they represent the 99th percentile of the distances of cis-eQTLs from their adjacent gene's transcript start and end sites. This is according to a comprehensive genome-wide analysis of putative functional regulatory elements (cis-eQTLs) using expression data from human lymphoblastoid cell lines [54]. These boundaries were chosen in attempt to capture association signals from proximal regulatory regions, in addition to the coding region. For the analysis of the DGI GWA study and the DGI permutations (used for method development purposes), ±50 kb was used, as these analyses were done before the Veyrieras et al. publication [54]. In the future, when transcriptional elements are comprehensively characterized for all genes in the genome, a discontinuous and more precise map of regulatory regions for each gene could be used for assigning SNPs to genes.
