No clear benefits, and in some cases significant decreases, were observed for local ancestry-informed weights compared to weights from a European or African ancestry GWAS or fixed-effects meta-analysis. Individuals with high, intermediate, and low European ancestry had lower accuracy using local ancestry-informed weights compared to the best weighting in each ancestry group: r = 0.71 versus 0.73 (from fixed-effect or European weights; p = 0.58); r = 0.61 versus 0.64 (from fixed-effect weights; p = 0.004); and r = 0.63 versus 0.65 (from African weights; p = 0.60), respectively (Figure 2).
