A list of 26,914 human gene transcripts was downloaded from the UCSC Genome Browser (http://genome.ucsc.edu/) in RefFlat format based on the human March 2006 (hg18) assembly. In the current study, 18,434 unique genes were used (17,680 on autosomes, and 754 on sex chromosomes), after filtering out genes with two or more transcripts that lie more than 1Mb apart on the same chromosome or that lie on separate chromosomes. All the genotyped or imputed SNPs that lie within an added physical distance upstream or downstream to a gene's most extreme transcript start and end sites of all its known splicing isoforms (intron and other non-coding sequences included) were assigned to each of the 18,434 genes. For gene set enrichment analysis of T2D, glycemic traits, and lipid and lipoprotein traits we used 110 kb upstream to the gene's most extreme transcript start 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
