Estimation of Breeding Values Using Different Densities of Snp to Inform Kinship in Broiler Chickens

Authors

Mayara Salvian, Universidade de São Paulo Escola Superior de Agricultura Luiz de Queiroz
Gabriel Costa Monteiro Moreira, Universidade de Sao Paulo Escola Superior de Agricultura Luiz de Queiroz
Ândrea Plotzki Reis, Universidade Federal de Vicosa
Brayan Dias Dauria, Universidade de Sao Paulo Escola Superior de Agricultura Luiz de Queiroz
Fabrício Pilonetto, Universidade de Sao Paulo Escola Superior de Agricultura Luiz de Queiroz
Izally Carvalho Gervásio, Universidade de Sao Paulo Escola Superior de Agricultura Luiz de Queiroz
Mônica Corrêa Ledur, Empresa Brasileira de Pesquisa Agropecuaria Suinos e Aves
Luiz Lehmann Coutinho, Universidade de Sao Paulo Escola Superior de Agricultura Luiz de Queiroz
Matthew L. Spangler, University of Nebraska - LincolnFollow
Gerson Barreto Mourao, University of Sao Paulo, BrazilFollow

Document Type

Article

Date of this Version

2023

Citation

Published in Livestock Science 267 (2023) 105124. doi:10.1016/j.livsci.2022.105124

Comments

Used by permission.

Abstract

Background: Traditionally, breeding values are estimated based on phenotypic and pedigree information using the numerator relationship (A) matrix. With the availability of genomic information, genome-wide markers can be included in the estimation of breeding values through genomic kinship. However, the density of genomic information used can impact the cost of implementation. The aim of this study was to compare the rank, accuracy, and bias of estimated breeding values (EBV) for organs [heart (HRT), liver (LIV), gizzard (GIZ), lungs (LUN)] and carcass [breast (BRST), drumstick (DRM) and thigh (THG)] weight traits in a broiler population using pedigree-based BLUP (PBLUP) and single-step genomic BLUP (ssGBLUP) methods using various densities of SNP and variants imputed from whole-genome sequence (WGS).

Results: For both PBLUP and ssGBLUP, heritability estimates varied from low (LUN) to high (fHRT, LIV, GIZ, BRST, DRM and THG.) Regression coefficients values of EBV on genomic estimated breeding values (GEBV) were similar for both the high density (HD) and WGS sets of SNPs ranging from 0.87 to 0.99 across senarios.

Conclusion: Results show no benefit of using WGS data compared to HD array data using an unweighted ssGBLUP. Our results suggest that 10% of the content of the HD array can yield unbiased and accurate EBV.