Figure 4: Imputation with the Japanese reference panel.
From: Rare variant discovery by deep whole-genome sequencing of 1,070 Japanese individuals
(a) Comparison of imputation performance (r2) for four reference panels: 1,070 individuals in 1KJPN (1KJPN), 1,092 cosmopolitan samples in 1KGP (1KGP ALL), 1KJPN plus 1KGP ALL (1KJPN+1KGP ALL) and 89 Japanese individuals in 1KGP (1KGP JPT). The x axis represents the MAF of each panel. The y axis represents the averaged r2 at SNV sites that exist in both the cosmopolitan samples of 1KGP and 1KJPN. (b) A Manhattan plot of P values from GWAS of MMD. The SNV sites from the original data set and imputed markers are plotted as dots in magenta and grey, respectively. Blue and red lines display the significance threshold of the original and imputed results, respectively. Only one significant signal was identified on chromosome 17. (c) A plot of P values from GWAS of MMD with the original (non-imputed; upper panel) and imputed (lower panel) data set around the SNP exhibiting the significant signal in b. In the imputed result, the SNP with the highest association is a nonsynonymous variant of RNF213, and was reported as one of the MMD-causing variants in the original study. In contrast, from the non-imputed result the SNP with the highest association is located in the coding region of ENDOV.
