Genotype imputation strategies for Portuguese Holstein cattle using different SNP panels.

• Autor(a) principal: SILVA, A. A.
• Data de Publicação: 2019
• Outros Autores: SILVA, F. F., SILVA, D. A., SILVA, H. T., COSTA, C. N., LOPES, P. S., VERONEZE, R., THOMPSON, G., CARVALHEIRA, J.
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da EMBRAPA (Repository Open Access to Scientific Information from EMBRAPA - Alice)
• Texto Completo: http://www.alice.cnptia.embrapa.br/alice/handle/doc/1117818; https://doi.org/10.17221/120/2019-CJAS
• topic: Genomic evaluation; Imputation accuracy; Dairy cattle

Although several studies have investigated the factors affecting imputation accuracy, most of these studies involved a large number of genotyped animals. Thus, results from these studies cannot be directly applied to small populations, since the population structure affects imputation accuracy. In addition, factors affecting imputation accuracy may also be intensified in small populations. Therefore, we aimed to compare different imputation strate-gies for the Portuguese Holstein cattle population considering several commercially available single nucleotide poly-morphism (SNP) panels in a relatively small number of genotyped animals. Data from 1359 genotyped animals were used to evaluate imputation in 7 different scenarios. In the S1 to S6 scenarios, imputations were performed from LDv1, 50Kv1, 57K, 77K, HDv3 and Ax58K panels to 50Kv2 panel. In these scenarios, the bulls in 50Kv2 were divided into reference (352) and validation (101) populations based on the year of birth. In the S7 scenario, the validation population consisted of 566 cows genotyped with the Ax58K panel with theirgenotypes masked to LDv1. In general, all sample imputation accuracies were high with correlations ranging from 0.94 to 0.99 and concordance rate rang-ing from 92.59 to 98.18%. SNP-specific accuracy was consistent with that of sample imputation. S4 (40.32% of SNPs imputed) had higher accuracy than S2 and S3, both with less than 7.59% of SNPs imputed. Most probably, this was due to the high number of imputed SNPs with minor allele frequency (MAF) < 0.05 in S2 and S3 (by 18.43% and 16.06% higher than in S4, respectively). Therefore, for these two scenarios, MAF was more relevant than the panel density. These results suggest that genotype imputation using several commercially available SNP panels is feasible for the Portuguese national genomic evaluation.