Bayesian model combining linkage and linkage disequilibrium analysis for low density-based genomic selection in animal breeding

• Published in: Journal of Applied Animal Research Vol. 46; no. 1; pp. 873 - 878
• Main Authors: Silva, Fabyano Fonseca, Jerez, Elcer Albenis Zamora, de Resende, Marcos Deon Vilela, Viana, José Marcelo Soriano, Azevedo, Camila Ferreira, Lopes, Paulo Sávio, Nascimento, Moysés, de Lima, Rodrigo Oliveira, Guimarães, Simone Eliza Facioni
• Format: Journal Article
• Language: English
• Published: Taylor & Francis 01-01-2018; Taylor & Francis Group
• Subjects: Genetic architecture; genome prediction; genotypic IBD matrix; SNP marker
• ISSN: 0971-2119; 0974-1844
• DOI: 10.1080/09712119.2017.1415903

We combined linkage (LA) and linkage disequilibrium (LDA) analyses (emerging the term 'LALDA') for genomic selection (GS) purposes. The models were fitted to a simulated dataset and to a real data of feed conversion ratio in pigs. Firstly, the significant QTLs (quantitative trait locus) were identified through LA-based mixed models considering the QTL-genotypes as random effects by means of genotypic identity by descent matrix. This matrix was calculated at the positions of significant QTLs (based on LA) allowing to include the QTL-genotype effects additionally to SNP (single nucleotide polymorphism) markers (based on LDA) and additive polygenic effects in several GS models (Bayesian Ridge Regression - BRR; Bayes A - BA; Bayes B - BB; Bayes C - BC and Bayesian LASSO - BL). These models combing all mentioned effects were denominated LALDA. Goodness-of-fit and predictive ability analyses were performed to evaluate the efficiency of these models. For the real data, although slightly, the superiority of the LALDA models was verified in comparison to traditional LDA models for GS. For the simulated dataset, the models presented similar results. For both LDA and LALDA frameworks, BA showed the best fitting through Deviance Information Criterion and higher predictive ability in the simulated and real datasets.