Exploiting whole genome sequence variants in cattle breeding : Unraveling the distribution of genetic variants and role of rare variants in genomic evaluation

Abstract

The availability of whole genome sequence data enables to better explore the genetic mechanisms underlying different quantitative traits that are targeted in animal breeding. This thesis presents different strategies and perspectives on utilization of whole genome sequence variants in cattle breeding. Using whole genome sequence variants, I show the genetic variation, recent and ancient inbreeding, and genome-wide pattern of introgression across the demographic and breeding history in different cattle populations. Using the latest genomic tools, I demonstrate that recent inbreeding can accurately be estimated by runs of homozygosity (ROH). This can further be utilized in breeding programs to control inbreeding in breeding programs. In chapter 2 and 4, by in-depth genomic analysis on whole genome sequence data, I demonstrate that the distribution of functional genetic variants in ROH regions and introgressed haplotypes was shaped by recent selective breeding in cattle populations. The contribution of whole genome sequence variants to the phenotypic variation partly depends on their allele frequencies. Common variants associated with different traits have been identified and explain a considerable proportion of the genetic variance. For example, common variants from whole genome sequence associated with longevity have been identified in chapter 5. However, the identified common variants cannot explain the full genetic variance, and rare variants might play an important role here. Rare variants may account for a large proportion of the whole genome sequence variants, but are often ignored in genomic evaluation, partly because of difficulty to identify associations between rare variants and phenotypes. I compared the powers of different gene-based association mapping methods that combine the rare variants within a gene using a simulation study. Those gene- based methods had a higher power for mapping rare variants compared with mixed linear models applying single marker tests that are commonly used for common variants. Moreover, I explored the role of rare and low-frequency variants in the variation of different complex traits and their impact on genomic prediction reliability. Rare and low-frequency variants contributed relatively more to variation for health-related traits than production traits, reflecting the potential of improving prediction reliability using rare and low-frequency variants for health-related traits. However, in practice, only marginal improvement was observed using selected rare and low-frequency variants when combined with 50k SNP genotype data on the reliability of genomic prediction for fertility, longevity and health traits. A simulation study did show that reliability of genomic prediction could be improved provided that causal rare and low-frequency variants affecting a trait are known.