Abstract:

【Background】 In the breeding work of livestock and poultry, the core of which is the accuracy of genomic estimated breeding values. Different levels of genetic marker densities have a great impact on estimated breeding values, and with the development of genotyping technology and the decrease of high-throughput sequencing prices, genomic selection studies based on sequencing data have emerged. Theoretically, higher marker density can obtain higher prediction accuracy. This is because Quantitative Trait Loci (QTL) affecting the target trait are in linkage disequilibrium with at least one of the high-density markers covering the entire genome. A higher density of marker levels theoretically ensures tight linkage between markers and QTL, thus ensuring higher prediction accuracy. However, compared with microarray data, it has also been shown that the accuracy of genomic prediction for imputed sequencing data is not significantly improved. 【Objective】 Using the GBLUP method, we compared the genomic selection accuracy of imputed sequencing data and microarray data for slaughter traits in broiler chickens to provide a theoretical basis for genotyping strategies for broiler genomic selection breeding. 【Method】 In this study, we used SNP array data and imputed whole-genome sequence level (WGS) data to perform genomic prediction for the traits of breast muscle weight, carcass weight and thigh muscle weight in white feather broilers using the GBLUP method, and then we conducted a comparative study on their accuracy in genomic prediction. First, 3 362 chickens were genotyped using the Jingxin No. 1 chicken 55 K SNP chip, and 230 chickens were randomly selected from the ninth batch of generation 7 for whole-genome resequencing, and then the 55 K SNP chip data were imputed to the resequencing data level using Beagle 5.1 software. Considering the effect of chromosome size on the filling accuracy, the larger chromosome 3 and the smaller chromosome 14 were used to calculate the allele correct rate (CR) and genotype correlation coefficient (Cor), and the imputed WGS accuracy was determined by this study. The genomic breeding values of three slaughter traits were predicted using the imputed WGS data, and the accuracy, rank correlation and unbiasedness of the prediction results were evaluated using a 5-fold cross-validation method. 【Result】 The results showed that the average allelic accuracy of the two chromosomes was 0.924 and the average genotype correlation was 0.885, and the imputed WGS accuracy was high enough to be used for genomic prediction studies at a later stage. The accuracy of the predicted genomic breeding values calculated from microarray data ranged from 0.2194 to 0.2629, and the accuracy of the predicted genomic breeding values calculated from imputed sequencing data ranged from 0.2110 to 0.2695. The results show that the difference in the accuracy of the prediction of genomic breeding values from the imputed sequencing data was not significant compared with the 55 K SNP chip results. 【Conclusion】 Compared with the results of 55 K SNP microarray, the improvement in the accuracy of genomic breeding value prediction for three slaughter traits (breast muscle weight, carcass weight and leg muscle) in white feather broiler using imputed genomic level data was not significant, which provides a reference for the selection of data types in livestock genetic breeding work.

Key words: white feather broiler, slaughter traits, genomic breeding value prediction, imputed sequencing data, microarray data, assessment