To set expectations and guide design of the full GTEx project, we built a framework to evaluate the statistical power to detect eQTLs. The statistical power depends on various parameters, some known more accurately than others. These parameters include the number of donors, the eQTL effect size, the noise, as well as the significance threshold, which is based on the number of hypotheses tested. Assuming we are testing cis-eQTLs between each of the 20,000 genes and 10 non-redundant cis-SNPs (on average) in vicinity (±100 kb) of each gene, the overall number of hypotheses is 200,000. Therefore, using a Bonferroni correction, we set the significance threshold, α, to be 0.05/200,000. For a trans-eQTL analysis, a conservative estimate of α is ~ 5×10−13 (20,000 transcripts tested against 5 million loci). We model the expression data as log-normally distributed with a log standard deviation of 0.13 within each genotype class. This level of noise is based on estimates from initial GTEx data. The effect size depends both on the minor allele frequency of the SNP (MAF) and the actual log expression change between
