Table 1. The accuracy and minor allele concordance rate of imputed genotypes.
Population | n | % of 86k SNPs that are polymorphic | % SNPs with any missing data | Imputation method (recomb. rate source) | Refer. haplo. | Mean r 2 by SNP | Mean minor allele concord. (%) |
---|---|---|---|---|---|---|---|
Thailand(THL) | 91 | 42% | 34% | IMPUTE2 | 90 THL | 0.51 | 31.8 |
(GB4x7G8) | 471 ALL | 0.70 | 66.5 | ||||
IMPUTE2 | 90 THL | 0.53 | 38.2 | ||||
(HB3xDD2) | 471 ALL | 0.70 | 67.2 | ||||
IMPUTE2 | 90 THL | 0.56 | 46.1 | ||||
(LDhat) | 471 ALL | 0.71 | 70.7 | ||||
BEAGLE | 90 THL | 0.92 | 91.4 | ||||
(map free) | 471 ALL | 0.27 | 29.0 | ||||
Cambodia (CBO) | 253 | 60% | 45% | IMPUTE2 | 252 CBO | 0.77 | 68.8 |
(GB4x7G8) | 471 ALL | 0.87 | 82.5 | ||||
IMPUTE2 | 252 CBO | 0.78 | 73.7 | ||||
(HB3xDD2) | 471 ALL | 0.88 | 83.2 | ||||
IMPUTE2 | 252 CBO | 0.84 | 79.1 | ||||
(LDhat) | 471 ALL | 0.90 | 86.1 | ||||
BEAGLE | 252 CBO | 0.96 | 92.1 | ||||
(map free) | 471 ALL | 0.65 | 63.3 | ||||
Gambia (GMB) | 55 | 38% | 38% | IMPUTE2 | 54 GMB | 0.99 | 95.4 |
(GB4x7G8) | 471 ALL | 0.99 | 95.2 | ||||
IMPUTE2 | 54 GMB | 0.99 | 95.4 | ||||
(HB3xDD2) | 471 ALL | 0.99 | 95.4 | ||||
IMPUTE2 | 54 GMB | 0.99 | 95.4 | ||||
(LDhat) | 471 ALL | 0.99 | 95.4 | ||||
BEAGLE | 54 GMB | 0.90 | 87.1 | ||||
(map free) | 471 ALL | 0.21 | 23.0 | ||||
Malawi (MLW) | 60 | 41% | 37% | IMPUTE2 | 59 MLW | 0.34 | 9.5 |
(GB4x7G8) | 471 ALL | 0.25 | 20.2 | ||||
IMPUTE2 | 59 MLW | 0.30 | 9.2 | ||||
(HB3xDD2) | 471 ALL | 0.25 | 20.8 | ||||
IMPUTE2 | 59 MLW | 0.69 | 49.0 | ||||
(LDhat) | 471 ALL | 0.26 | 12.3 | ||||
BEAGLE | 59 MLW | 0.87 | 85.0 | ||||
(map free) | 471 ALL | 0.06 | 6.7 |
Accuracy is consistently high with Beagle when only population-specific haplotypes are used as references; improves with IMPUTE when a cosmopolitan reference panel is used; and is highest with IMPUTE when LDhat-inferred recombination rates are used. ALL refers to a cosmopolitan panel including populations from the listed countries (n = 459) and Vietnam (n = 12).