any of the five sets at a 10% FDR was only 20, and the mean detection rate over all sets was 12. This demonstrates well the biological utility of the eSNP set, given that this set is significantly enriched for SNPs that associate with expression traits, beyond the initial set of expression traits that defined the eSNP set, compared to comparable sets of randomly selected SNPs. A number of eQTL hot spots emerged as well in this full set of expression traits, where a given locus was identified as a hot spot if greater than 20 expression traits linked to a single eSNP at the locus. The significance of these hot spots was assessed by permuting the genotypes and examining the distribution of associations in the permuted sets. In each permutation set, we identified the maximum number of traits associated with a single marker over all markers. The mean of the maximum counts over 10 permutation sets was only 12, compared to a maximum of 283 in the observed data (Table S2).