Estratégias de imputação e associação genômica com dados de sequenciamento para características de produção de leite na raça Gir
Carregando...
Data
2018-02-22
Autores
Nascimento, Guilherme Batista do [UNESP]
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
A implementação de dados de sequenciamento de nova geração - “next-generation sequence” (NGS) em programas de melhoramento genético animal representa a mais recente ferramenta na utilização de dados genotípicos nos modelos de associação genômica, tendo em vista que todo polimorfismo é considerado nas associações entre registros fenotípicos e dados de sequenciamento. Como em toda nova tecnologia, a prospecção das variantes ainda representa um desafio no sentido computacional e de viabilidade dos custos para sua implementação em larga escala. Diante desses desafios, neste trabalho buscou-se meios de explorar os benefícios na utilização da NGS nas predições genômicas e superar as limitações inerentes a esse processo. Registros fenotípicos e genotípicos (Illumina Bovine HD BeadChip) de 2.279 animais da raça Gir (Bos taurus indicus) foram disponibilizados pela Embrapa Gado de Leite (MG) e utilizados para as análises de associação genômica. Além disso, dados de sequenciamento de 53 animais do 1000 “Bulls Project” deram origem à população de referência de imputação. Visando verificar a eficiência de imputação, foram testados diferentes cenários quanto a sua acurácia de imputação por meio da análise “leave-one-out”, utilizando apenas os dados de sequenciamento, que apresentaram eficiências de até 84%, no cenário com todos os 51 animais disponíveis após o controle de qualidade. Também foram verificadas as influências das variantes em baixa frequência na acurácia de imputação em diferentes regiões do genoma. Com a escolha da melhor estrutura da população de referência de imputação e aplicação dos controles de qualidade nos dados de NGS e genômicos, foi possível imputar os 2.237 animais genotipados, que passaram pelo controle de qualidade para dados de sequenciamento e realizar análise de associação genômica para as características produção de leite (PL305), teor de gordura (PG305), proteína (PP305) e sólidos totais (PS305), mensuradas aos 305 dias em animais da raça Gir leiteiro. Para tal, foram utilizados os valores genéticos desregredidos (dEBV) como variável resposta no modelo de regressão múltipla. Regiões de 1Mb que contivessem 100 ou mais variantes com “False Discovery Rate” (FDR) inferior a 0,05, foram consideradas significativas e submetidas a análise de enriquecimento por meio dos termos MeSh (“Medical Subject Headings”). As três regiões significativas (FDR<0,05) para PS305 foram observadas nos cromossomos 11, 12 e 28 e a única região significativa em PG305 foi no cromossomo 6. Tais regiões apresentaram variantes associadas com vias metabólicas da produção de leite, ausentes nos painéis comerciais de genotipagem, podendo representar genes candidatos a seleção.
- Implementing "next-generation sequence" (NGS) data in animal breeding programs represents the latest tool in the use of genotypic data in genomic association models, since all polymorphisms are considered in the associations between phenotypic records and sequencing data. As with any new technology, variant prospecting still represents a computational and cost-effective challenge for large-scale implementation. Front to these challenges, this work sought ways to explore the benefits of using NGS in genomic predictions and overcome the inherent limitations of this process. Phenotypic and genotypic (Illumina Bovine HD BeadChip) records of 2,279 Gir animals (Bos taurus indicus) were made available by Embrapa Gado de Leite (MG) and used for genomic association analysis. In addition, sequence data of 53 animals from the 1000 Bulls Project gave rise to the imputation reference population. In order to verify the imputation efficiency, different scenarios were tested for their imputation accuracy through the leave-one-out analysis, using only the sequencing data, which presented efficiencies of up to 84%, in the scenario with all the 51 animals available after quality control. Influences from the low-frequency variants on the accuracy of imputation in different regions of the genome were also verified. After identifying the best reference population structure of imputation and applying the quality controls in the NGS and genomic data, it was possible to impute the 2 237 genotyped animals that passed in the quality control to sequencing data and perform genomic association analysis for (PL305), fat content (PG305), protein (PP305) and total solids (PS305), measured at 305 days in dairy Gir animals. For this, unregulated genetic values (dEBV) were used as response variable in the multiple regression model. Regions of 1Mb containing 100 or more variants with a False Discovery Rate (FDR) lower than 0.05 were considered statistically significant and submitted to pathways enrichment analysis using the MeSh (Medical Subject Headings) terms. The three significant regions (FDR <0.05) for PS305 were observed on chromosomes 11, 12 and 28 and only one significant region in PG305, was on chromosome 6. These regions presented variants associated with metabolic pathways of milk production, absent in the panels genotyping, and may represent genes that are candidates for selection
- Implementing "next-generation sequence" (NGS) data in animal breeding programs represents the latest tool in the use of genotypic data in genomic association models, since all polymorphisms are considered in the associations between phenotypic records and sequencing data. As with any new technology, variant prospecting still represents a computational and cost-effective challenge for large-scale implementation. Front to these challenges, this work sought ways to explore the benefits of using NGS in genomic predictions and overcome the inherent limitations of this process. Phenotypic and genotypic (Illumina Bovine HD BeadChip) records of 2,279 Gir animals (Bos taurus indicus) were made available by Embrapa Gado de Leite (MG) and used for genomic association analysis. In addition, sequence data of 53 animals from the 1000 Bulls Project gave rise to the imputation reference population. In order to verify the imputation efficiency, different scenarios were tested for their imputation accuracy through the leave-one-out analysis, using only the sequencing data, which presented efficiencies of up to 84%, in the scenario with all the 51 animals available after quality control. Influences from the low-frequency variants on the accuracy of imputation in different regions of the genome were also verified. After identifying the best reference population structure of imputation and applying the quality controls in the NGS and genomic data, it was possible to impute the 2 237 genotyped animals that passed in the quality control to sequencing data and perform genomic association analysis for (PL305), fat content (PG305), protein (PP305) and total solids (PS305), measured at 305 days in dairy Gir animals. For this, unregulated genetic values (dEBV) were used as response variable in the multiple regression model. Regions of 1Mb containing 100 or more variants with a False Discovery Rate (FDR) lower than 0.05 were considered statistically significant and submitted to pathways enrichment analysis using the MeSh (Medical Subject Headings) terms. The three significant regions (FDR <0.05) for PS305 were observed on chromosomes 11, 12 and 28 and only one significant region in PG305, was on chromosome 6. These regions presented variants associated with metabolic pathways of milk production, absent in the panels genotyping, and may represent genes that are candidates for selection
Descrição
Palavras-chave
Acurácia de imputação, False Discovery Rate, Next-generation sequence, Bovinocultura de leite, Accuracy of imputation, Dairy Cattle