Abstract: The study aimed to explore the efficiency of low-density liquid-based chips in practical animal breeding and reduce breeding costs. A total of 3 761 healthy Large White pigs with about 160 days of age and 110 kg of body weight were used in this study, 100 out of these Large White pigs were randomly selected to serve as a validation population for imputation, in which a 10K SNP panel was generated by sampling markers from the 50 K panel, according to the marker information in 10K SNP panel. Meanwhile, 800, 2 000, 3 600 individuals were randomly selected from the remaining pigs as the reference populations, respectively. Beagle 4.1 was used to impute the 100 validation pigs to 50K SNP panel, and 10 replications were carried out. Genotype consistency and genotype correlation coefficients were used to evaluate the accuracy of genotype imputation. The results showed that the average linkage disequilibrium (r²) of the 10K and 50K SNP panel were 0.227 and 0.258, respectively, with little difference. Minor allele frequency (MAF) of 0.05 was the inflection point for the accuracy of genotype imputation, the accuracy significantly increased after removing markers with MAF less than 0.05. With the reference population size increased from 800 to 3 600, the imputation accuracy was improved from 0.90 to 0.95, and the standard deviation of imputation accuracy among 10 replicates was also decreased from 0.006 to 0.002. For small reference population with 800 individuals, the genotype imputation accuracy of each chromosome fluctuated obviously, while with enlarging the reference population, the imputation accuracy of each chromosome became very close. The study result demonstrates that the imputing from 10K to 50K liquid-based SNP panel is feasible and could be helpful for the application in genomic selection on a large scale, dramatically reducing the breeding cost of genomic selection.

Key words: pig, liquid chip, genotype imputation, molecular breeding