Estudo da acurácia de imputação em bovinos da raça Canchim

Abstract

Imputation methods are applied to predict genotypes in a target population genotyped with lower density panels using genotypes from reference population genotyped with higher density panels, in order to reduce genotyping costs and usage in genomic selection. Thus, the aim of this study was to evaluate potential benefits of incorporating founder breed genotypes (Nelore and Charolais) for imputation in the Canchim composite breed. Genotypes from 285 Canchim (CA) and 114 MA genetic group (MA) individuals were used, being called as target population. The reference populations consisted of 814 Nelore (NE) and 897 Charolais (CH) animals, and in some cases, CA or MA animals were used. The panels considered were: BovineHD - BeadChip Illumina with 777,962 markers (reference population) and BovineSNP50 BeadChip Illumina with 54,609 markers (target population). A total of 17 scenarios were created to analyze the accuracy of imputation through concordance rate and r 2 allelic. The variation observed between the imputation scenarios using the r² allelic ranged from 0.59 (NE x MA) to 0.93 (CH+CA x MA). The integration of genotypes from Charolais resulted in concordance rates that ranged from 70.93% to 73.61%, being higher than the scenarios in which Nelore were used (varying from 63.24% to 65.42%). When combining the founders breed into the reference population, concordance rates ranging from 83.47% to 85.99% were observed; however, these scenarios presented half the quantity of markers in the reference populations when compared to the previous scenarios. The incorporation of founders breed genotypes increased the imputation accuracy for CA and MA. The inclusion of the Charolais as a reference population together with Canchim promoted the greatest gains in accuracy observed. It can be inferred that the use of multibreed reference populations may assist to reduce genotyping costs in composite breeds.