A PRS for an individual is a summation of their genotypes at variants genome-wide, weighted by effect sizes on a trait of interest. Effect sizes are typically estimated from published GWAS results, and only variants exceeding a P-value threshold, PT, are included (Dudbridge, 2013). Since even large GWAS achieve only marginal evidence for association for many causal variants, PRS are usually calculated at a set of P-value thresholds, e.g. PT=1×10−5,1×10−4,…,0.05,0.1,…,0.5. A common application of PRS is to test for shared genetic aetiology between traits. Here PRS on the base phenotype are calculated, using GWAS results, in individuals from an independent data set, and these are used as predictors of the target phenotype in a regression (see Supplementary Note S1). This technique was first applied by the International Schizophrenia Consortium (2009), demonstrating that genetic risk for schizophrenia (SCZ) is a predictor of bipolar disorder. This study also acted as a proof-of-principle for PRS, showing that PRS based on thousands of common variants genome-wide, including many with no effect and using effect size estimates from published GWAS, can provide a reliable indicator
