Abstract
ABSTRACT Genome-wide selection (GWS) uses simultaneously the effect of the thousands markers covering the entire genome to predict genomic breeding values for individuals under selection. The possible benefits of GWS are the reduction of the breeding cycle, increase in gains per unit of time, and decrease of costs. However, the success of the GWS is dependent on the choice of the method to predict the effects of markers. Thus, the objective of this work was to predict genomic breeding values (GEBV) through artificial neural networks (ANN), based on the estimation of the effect of the markers, compared to the Ridge Regression-Best Linear Unbiased Predictor/Genome Wide Selection (RR-BLUP/GWS). Simulations were performed by software R to provide correlations concerning ANN and RR-BLUP/GWS. The prediction methods were evaluated using correlations between phenotypic and genotypic values and predicted GEBV. The results showed the superiority of the ANN in predicting GEBV in simulations with higher and lower marker densities, with higher levels of linkage disequilibrium and heritability.
Highlights
Genome-wide selection (GWS) consists of using hundreds to thousands of markers saturating the genome in order to predict genomic breeding values (GEBV) of the individuals through statistical methods based on the estimation of effects of markers
RR-Best Linear Unbiased Prediction (BLUP) uses the same estimator as the ridge regression, but estimates the parameter penalized by the Restricted Maximum Likelihood (REML) (SCHULZ-STREECK; OGUTU; PIEPHO, 2011)
The choosing of the scenarios was based on a given character that presents any of these characteristics; knowing in which there would be greater or lesser correlations between the phenotypic and predicted values; and comparing these same scenarios applied with other statistical methods (RESENDE et al, 2012)
Summary
Genome-wide selection (GWS) consists of using hundreds to thousands of markers saturating the genome in order to predict genomic breeding values (GEBV) of the individuals through statistical methods based on the estimation of effects of markers. Unlike marker-assisted recurrent selection, uses all estimated effects of marker loci regardless of whether or not they are significantly associated with the phenotype in order to predict the GEBV of each individual in the population under selection (SINGH; SINGH, 2015). The statistical models employed to predict GEBV provide different assumptions on the number and effects of Quantitative Trait Loci (QTL). The methods differ in general in the assumption on the genetic model associated with the quantitative character (RESENDE et al, 2008). In the RR-BLUP method, the effects of all markers are estimated simultaneously and are assumed as random and with the same allelic frequency, contributing to genetic variation (RESENDE et al, 2011)
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
