Skip to main content
. Author manuscript; available in PMC: 2018 Dec 1.
Published in final edited form as: Genet Epidemiol. 2017 Nov 7;41(8):811–823. doi: 10.1002/gepi.22083

Table 5. Accuracy of main prediction methods in analyses of height in an African cohort.

Model Weights associated to each predictor Adjusted R2 P-value for improvement over simpler model
European PRS African PRS
EUR 0.164 0.02618 <10-11
AFR 0.106 0.01074 <10-5
AFR+ANC 0.124 0.01331 0.01
EUR+AFR 0.155 0.092 0.03397 <10-3
EUR+AFR+ANC 0.150 0.102 0.03443 0.17
EUR-AFR-meta 0.151 0.027 0.03158 NA

We report adjusted R2 on the observed scale for each of the 5 main prediction methods, as well as EUR-AFR-meta. P-values are from likelihood ratio tests comparing models EUR and AFR to the null model, model AFR+ANC to AFR, model EUR+AFR to EUR, and EUR+LAT+ANC to EUR+AFR. For the EUR model we used RLD2=0.2 and PT=10-3, for AFR we used RLD2=0.8 and PT=0.05 and for EUR-AFR-meta we used RLD2=0.2and PT=10-6. We also report normalized weights, defined as the mixing weight α^k (see Methods) multiplied by the standard deviation of the PRS.