The availability of whole genome expression and SNP data allows the elucidation of genetic effects acting in trans. The number of 2.2 million SNPs per population is large (MAF > 0.05) so testing all SNPs against all genes becomes computationally and statistically challenging (correcting for millions of tests). We took a “candidate variant approach” by testing only putatively functional SNPs. The goal of the analysis is not to compare the numbers of cis vs. trans effects, which is an irrelevant question in the genome-wide context especially given differential power in detection. The goal is to assess the relative contribution of primary molecular variants in trans. Towards this end, we selected four categories of SNPs to analyze for trans effects: i) SNPs with functional effects on gene expression in cis (as determined above in the single population analyses), ii) non-synonymous SNPs (Ensembl v41 annotation); iii) SNPs influencing splicing (Ensembl v41 annotation) iv) and SNPs within microRNAs (as annotated in miRBase). These correspond to approximately 25,000 SNPs (MAF > 0.05) per population (see Table 3 for counts in each category).
