Figure 2

Scatterplots of the cumulative distribution function of $Y_1\left(F_1\right)$ versus $Y_2\left(F_2\right)$ without conditioning on any variant (left panel) and conditional on the variant at 41,984,243 base-pair position (right panel). The indicated blue dots denote individuals with high adjusted phenotypes having 1 minor allele of the variant (no individual has 2 alleles). Without conditioning on the variant, these dots in the upper tail are closely clustered (left panel), but when the effect of the variant is removed (right panel), they spread out. Note that the unexplained association is likely to be caused by other unknown variants. If we could have identified all the variants that explain the association between phenotypes, then the plot in the right panel conditioning on these variants would have all dots randomly scattered on the unit square.