Applications of whole genome sequence data in beef cattle genomics

Over the last decade, genome sequencing technologies to obtain whole genome sequence (WGS) have improved a lot as they become more and more accessible and affordable to researchers in academia and industry. WGS provides an exciting opportunity to make use of rare variants and previously unidentified quantitative trait loci for various applications including GWAS and genomic prediction. However, WGS also comes with its own limitations beside the high cost of sequencing, which include high storage and memory requirements, and statistical and computational challenges related to high dimensional data. Therefore, there is a need to explore how WGS can be effectively utilized in beef cattle genomics. In this dissertation, I have analyzed simulated and real genomic data from various beef cattle breeds obtained using SNP chip technology as well as sequencing platforms to answer various questions on this end. First, I simulated matings of three beef cattle breeds (Angus, Hereford and Brahman) and their crosses over 20 generations to study the genomic prediction trends over time using dense SNP information either with or without known QTLs in the dataset. I demonstrate that while WGS may result in a minimal gain in prediction accuracy at a certain snapshot in time, collection of WGS data over the years will increase the accuracy of genomic prediction in beef cattle for distantly related animals. Second, I demonstrate that WGS imputation with a small reference population closely related to the target population is more accurate than a large multi-breed reference with distantly related animals. This implies that even though a large variety of dairy and beef breeds from around the world have already been sequenced, there is still a need to increase the representation of less European-centric breeds in sequencing efforts to improve the imputation of these breeds. Third, I observed that association analysis based on WGS resulted in stronger signals of association as compared to medium and high-density SNP chip genotypes, and identified novel loci regulating carcass traits. However, we observed no gain in prediction accuracy using imputed WGS as compared to medium and high-density genotype for genomic prediction in Korean beef cattle population using GBLUP methodology. We also observed that addition of SNPs pre-selected based on association analysis to medium and high-density genotypes may result in a slight increase in accuracy of prediction for traits regulated by large effect loci. Moreover, modelling epistatic effects explained a small proportion of genetic variance but failed to increase the prediction accuracy for additive genetic value and total genetic value. Lastly, I identified genomic selection signals in Angus and Hanwoo beef cattle by four distinct methods primarily based on allele frequency and haplotype structures surrounding SNP variants. We also combined these selection signals to obtain strong evidence of selection by composite selection signature score. Finally, we identified and confirmed various positional candidate genes located in selected regions related to beef production and quality. In conclusion, I have explored various strategies for effective utilization of WGS and identified novel genomic variants regulating meat traits in Hanwoo and Angus beef cattle. These analyses will help in effective application of WGS in beef cattle genomics.