Thread by @thatdnaguy

#ASHG21 RB: Richard Border.
Widespread evidence of systematic bias in estimates of genetic correlation due to assortative mating.
[ I think this is the preprint. Don't hold me to it. ]
biorxiv.org/content/10.110…

#ASHG21 RB: interested in characterizing what extent two traits share similarity across traits.

#ASHG21 RB: Can compute the polygenic scores for each trait and see how they compare. Or correlation of effect correlations. Many ways to look at genetic correlations.

#ASHG21 RB: We'll use effect correlation and score correlation. Expect them to be essentially the same thing. But this correlation depends on random mating. Can have trouble if mates are chosen due to phenotypic similarity. [ extremely important point ]

#ASHG21 RB: Examples include height, years of education. Across traits anxiety is cross correlated with schizophrenia and some others.

#ASHG21 RB: Knowing the spousal correlation you can do a decent job predicting genetic correlation.

#ASHG21 RB: true polygenic scores for different traits can become correlated due to similar phenotypic mate selection even if molecularly they shouldn't be correlated.