Fig. 4
Assessment of imputation performance and technical concordance across diverse populations. a Imputed genotypes calculated using lcWGS data were highly correlated with genotypes from known 1KGP data (n = 116), with all samples having an imputation quality above 0.90 r2. The thin gray dashed line demonstrates previously reported imputation quality from a genotyping array (r2 = 0.90) [4]. b GPSCAD calculated using lcWGS data was highly correlated (r2 = 0.98) with those calculated using known 1KGP data (n = 116). c GPSBC calculated using lcWGS data was highly correlated (r2 = 0.91) with those calculated using known 1KGP data (n = 116). d GPSAF was highly correlated (r2 = 0.98) with those calculated using known 1KGP data (n = 116). 1KGP, 1000 Genomes Project; lcWGS, low coverage whole genome sequencing; GPS, genome-wide polygenic score; CAD, coronary artery disease; BC, breast cancer; AF, atrial fibrillation