While the genetic correlation analysis provides a summary measure of the genomewide correlation of SNP effect sizes, the bivariate MiXeR analysis estimates the number of overlapping SNPs/loci between the two traitsregardless of the direction of effects (i.e. both concordant and discordant effects) and calculates the fraction of concordant variants. For instance, a genetic correlation of zero can be seen for traits that have no overlapping risk SNPs as well as for traits that have completely overlapping risk SNPs with 50% concordant and 50%discordant effect directions. So, as the huge overlap observed between depression and educational attainment (Supplementary TableS6A and Figure S9-1) is based on a majority of discordant variants (58%) (and fewer concordant variants (42%)), it is consistent with the overall negative correlation observed between the two traits.
