Assessing accuracy of imputation using different SNP panel densities in a multi-breed sheep population

Table 6 Accuracy of genotype imputation with the FIMPUTE software using two types of reference population

Scenario^a	CRAll^c	r²All^c	CRW^d	r²W^d	MeanA^e	MinA^e	MaxA^e	MeanW^e	MinW^e	MaxW^e
25_5K50K	93.39	89.38	89.16	82.17	0.023	0.079	0.467	0.145	0.049	0.376
26_5K50K	95.45	92.10	82.05	70.47	0.267	0.096	0.432	0.185	0.077	0.355
27_5K50K^b	89.07	82.06	–	–	0.180	0.055	0.401	–	–	–
28_5K50K	89.94	84.01	89.80	83.27	0.250	0.100	0.427	0.200	0.050	0.384
29_5K50K	96.24	93.12	87.55	79.76	0.283	0.085	0.426	0.201	0.075	0.387
30_5K50K	87.89	81.23	88.32	80.55	0.215	0.100	0.535	0.162	0.061	0.310
31_5K50K	90.16	82.17	65.05	41.57	0.243	0.109	0.413	0.03	0.001	0.260

^aGenotype imputation was from 5K to 50K using two types of reference population: (i) fixed reference population containing a large number of animals from all breeds and (ii) within-group reference population
^bScenario defined for the calculation of SNP r² using 1000 animals as imputed
^cCRAll and r²All = concordance rate and squared Pearson correlation, respectively, using the FIMPUTE software when a large set of animals from all breeds was defined as the reference population
^dCRW and r²W = concordance rate and squared Pearson correlation, respectively, using the FIMPUTE software when the within-group population was defined as the reference population
^eMeanA, MinA, MaxA, MeanW, MinW and MaxW = mean, min and max relationship among the 10 most related animals between the reference and imputed sets (all animals (A) or within-group (W))

ISSN: 1297-9686