|CARVALHEIRO, ROBERTO - Universidade Estadual Paulista (UNESP)|
|BOISON, ANA - University Of Natural Resources & Applied Life Sciences - Austria|
|NEVES, HAROLDO - Universidade Estadual Paulista (UNESP)|
|SARGOLZAEI, MEHDI - McGill University - Canada|
|SCHENKEL, FLAVIO - McGill University - Canada|
|UTSUNOMIYA, YURI - Sao Paulo State University (UNESP)|
|O'BRIEN, ANA M. PEREZ - University Of Natural Resources & Applied Life Sciences - Austria|
|SOLKNER, JOHANN - University Of Natural Resources & Applied Life Sciences - Austria|
|MCEWAN, JOHN - Agresearch|
|Van Tassell, Curtis - Curt|
|GARCIA, FERNANDO - Sao Paulo State University (UNESP)|
Submitted to: Genetics Selection Evolution
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/19/2014
Publication Date: 10/10/2014
Citation: Carvalheiro, R., Boison, A., Neves, H.H., Sargolzaei, M., Schenkel, F., Utsunomiya, Y.T., O'Brien, A., Solkner, J., Mcewan, J., Van Tassell, C.P., Sonstegard, T.S., Garcia, F. 2014. Genotype imputation efficiency in Nelore Cattle. Genetics Selection Evolution. 46(1):69.
Interpretive Summary: The purpose of this study was to define the best type of low density DNA assay (<100,000 DNA markers) for genotyping animals of lower economic value, and still have the highest accuracy for predicting the genotype information of a higher density (HD) DNA assay (>500,000 DNA markers) that is needed for predicting genetic merit. Often HD assay are cost prohibitive for most beef producers to generate a reasonable profit margin from breeding animals. The results indicated that if the HD is considered as the target chip for genomic applications in the Nelore breed, their cost effectiveness can be improved with the strategy of genotyping part of the economically marginal animals with a lower density chip containing around 15K useful SNPs and predicting the other 400,000 missing HD genotypes with an accuracy of 96%. A denser LD chip (50K useful SNPs) is recommended for animals poorly related within the reference population of a Zebu breed. This study also showed a preferential result for a specific software program for genotype prediction (FImpute over BEAGLE). Overall, these results suggest that cheaper, low density assays are feasible for genotyping unproven animals of unknown genetic value.
Technical Abstract: Genotype imputation efficiency in Nelore cattle was evaluated in different scenarios of lower density (LD) chips, imputation methods and sets of animals to have their genotypes imputed. Twelve commercial and virtual custom LD chips with densities varying from 7K to 75K SNPs were tested. Customized LD chips were designed taking into account minor allele frequency (MAF), linkage disequilibrium and distance among SNPs. The imputation methods adopted were FImpute and BEAGLE. From a total of 995 bulls and 1,247 cows genotyped with the Illumina Bovine HD chip (HD), 793 sires were used as the reference set, and the remaining 202 younger sires and all the cows were used as two separate imputed sets, which had their genotypes masked except for those SNPs present in the LD chip to be tested. The correlation between imputed and observed genotypes (CORR), assigned as 0 (AA), 1 (AB) or 2 (BB), was used as the main measure to assess imputation efficiency. Imputation efficiency increased as the proportion of SNPs to be imputed decreased. However, the gain of using LD chips with more than 15K SNPs was relatively small because efficiency was already high at this density. For the customized chips, a slight difference in favor of selecting SNPs based on MAF rather than on linkage disequilibrium was observed. The best result was obtained when both criteria, together with distance among SNPs, were combined to define the SNP content in acustom LD chip. Commercial and customized LD chips with equivalent densities presented similar results. FImpute outperformed BEAGLE for the different LD chips and imputed sets. There was a trend of higher difference between both methods for LD chips with lower densities. Regardless of the method, the CORR values tended to be higher as the relatedness between imputed and reference animals increased, especially for a 7K chip. In lower than average levels of relatedness, the dispersion of CORR values was higher for BEAGLE than for FImpute. The results indicated that if the HD is considered as the target chip for genomic applications in the Nelore breed, their cost effectiveness can be improved with the strategy of genotyping part of the animals with a chip containing around 15K useful SNPs and imputing their HD missing genotypes. For the studied population, FImpute should be preferred over BEAGLE to impute missing genotypes.