Assessment of genotype imputation methods

BMC Proceedings

Table 1 Mean error rates by imputation method and scenario

		IMPUTE	MACH	PLINK	fastPHASE
Scenario 1: Imputation of untyped SNPs
Overall		0.112	0.114	0.142	0.135
By region^a	null1	0.251	0.251	0.284	0.271
	null2	0.066	0.066	0.090	0.085
	PADI4-1	0.083	0.092	0.131	0.111
	PADI4-2	0.106	0.109	0.144	0.162
	PTPN22-1	0.099	0.098	0.122	0.107
	PTPN22-2	0.061	0.059	0.069	0.058
By max pairwise	r² < 0.5	0.208	0.212	0.245	0.248
LD	r² ≥ 0.5	0.030	0.030	0.053	0.038
Scenario 2: Imputation to combine two datasets
Overall		0.116	0.112	0.173	0.127
By region^a	null1	0.206	0.201	0.250	0.218
	null2	0.123	0.122	0.175	0.139
	PADI4	0.079	0.069	0.145	0.097
	PTPN22	0.055	0.053	0.121	0.053
By max pairwise	r² < 0.5	0.200	0.197	0.256	0.211
LD	r² ≥ 0.5	0.046	0.041	0.105	0.059

^aIn Scenario 1, for regions PADI4-1 and PTPN22-1, the most strongly associated SNP was removed and imputed, while for regions PADI4-2 and PTPN22-2, the two SNPs flanking the most strongly associated SNP were imputed, in addition to other SNPs as described in the methods.

ISSN: 1753-6561