Imputation and genomic studies in bovine Nelore

Author(s)

Priscila Arrigucci Bernardes

• Document type: Doctoral Thesis
• Press: Jaboticabal. 2018-07-26.
• Institution: Universidade Estadual Paulista (Unesp). Faculdade de Ciências Agrárias e Veterinárias. Jaboticabal
• Defense date: 2018-06-29
• Advisor: Danísio Prado Munari; Ricardo Vieira Ventura

Abstract

Among all the information provided by methodologies that use single nucleotide polymorphism (SNPs), the runs of homozygosity (ROH) and linkage disequilibrium have been used for studies that explore genomic information in beef cattle population, as the genome-wide association, genomic selection, the structure of population and others. Nowadays, the imputation is used in these studies to reduce genomic costs and this also can be used combining genomic information from different panels. The animals used to be imputed should present genotypes inferred with high accuracy to allow the use imputed genotypes in other studies. However, the accuracy is verified only if there is a real genotype to evaluate the imputed genotype. Therefore, this study aimed: (1) Evaluate imputation of commercial and customized low density panels to high density panels (Illumina and Affymetrix), as well as to a combined panel (Illumina + Affymetrix) in Nelore beef cattle, and estimating linkage disequilibrium and haplotype blocks conformation to high density panels individually and after imputation; (2) Study a strategy to predict imputation accuracy using artificial neural network and linear regression; (3) Study runs of homozygosity and inbreeding in a populations from Nelore beef cattle, as well as identify genes present in ROH with high frequency in population. For ROH studies were used 34 bulls from different lines and the progeny, totalizing 809 Nelore animals genotyped with information of 509.107 SNPs (Illumina). The imputation analysis and imputation accuracy prediction used the same animals, wherein 93 were also genotyped with Axion Genome-Wide BOS 1 Array Plate. The imputation analysis demonstrates that the combined panels used from different panels can be considered due to increasing the density and number of haplotype blocks, increasing the probability to find a marker close to an important QTL. Furthermore, this strategy indicates that the choice for common SNPs between high-density panels Illumina and Affymetrix to customize a lower density panel can increase the imputation accuracy to Illumina and Affymetrix. The prediction of imputation accuracy analysis showed that the neural network is more efficient compared to linear regression, and could be used for this purpose. The results from ROH analysis showed low population inbreeding, however the sires presented higher inbreeding compared to progenies and longer runs of homozigosity, which suggest that has occurred intense use of few sires in recent generations in some families. (AU)

• FAPESP's process: 15/25096-6 - Imputation, linkage desequilibrium, haplotype blocks and inbreeding in Nelore cattle
• Grantee: Priscila Arrigucci Bernardes
• Support Opportunities: Scholarships in Brazil - Doctorate