To answer these questions, we masked and imputed 40,300 SNPs from the Illumina 650Y array in a set of 1216 Gambian individuals from MalariaGEN. We repeated the analysis for each of four reference panels: all 2022 HapMap 3 haplotypes (HM3); 200 Gambian haplotypes (GMB); 200 Gambian haplotypes plus 200 Ghanaian haplotypes (GMB+GHN); and all of the aforementioned panels combined (GMB+GHN+HM3; 2422 haplotypes). For each imputation run, we used IMPUTE2 with khap = 500. In reference panels with fewer than 500 haplotypes, we reduced khap to the number of available haplotypes. The results are shown in Figure 3, which breaks down the imputation accuracy by the minor allele frequencies of the SNPs in the 1216 imputed Gambians.
