Neural networks for predicting breeding values and genetic gains

Gabi Nunes Silva,Moysés Nascimento,Rafael Simões Tomaz,Cosme Damião Cruz,Isabela de Castro Sant'Anna,Leonardo Lopes Bhering

doi:10.1590/0103-9016-2014-0057

Abstract

Analysis using Artificial Neural Networks has been described as an approach in the decision-making process that, although incipient, has been reported as presenting high potential for use in animal and plant breeding. In this study, we introduce the procedure of using the expanded data set for training the network. Wealso proposed using statistical parameters to estimate the breeding value of genotypes in simulated scenarios, in addition to the mean phenotypic value in a feed-forward back propagation multilayer perceptron network. After evaluating artificial neural network configurations, our results showed its superiority to estimates based on linear models, as well as its applicability in the genetic value prediction process. The results further indicated the good generalization performance of the neural network model in several additional validation experiments.

Highlights

Genetic breeding is a science aimed at increasing the frequency of alleles and/or obtaining favorable genotypic combinations, in order to increase the production efficiency of an individual or population
After evaluating artificial neural network configurations, our results showed its superiority to estimates based on linear models, as well as its applicability in the genetic value prediction process
In conventional statistical procedures used in the genetic gain prediction process, only mean genotypic values are considered assuming the existence of their correlation with the actual genetic values of the genotypes, denoted by heritability of the trait

Summary

Introduction

Genetic breeding is a science aimed at increasing the frequency of alleles and/or obtaining favorable genotypic combinations, in order to increase the production efficiency of an individual or population. A new paradigm can be employed in genetic breeding for selection purposes that does not involve stochastic modeling, but instead the principles of learning in a computational intelligence approach In this context, approaches based on Artificial Neural Networks (ANN) have been described as an additional tool in the decision making process in various fields of science with great potential in animal and plant genetics (Gianola et al, 2011; Nascimento et al, 2013; Ventura et al, 2012). An alternative is to use a virtual data set, provided by statistical techniques able to preserve some features of the original experiment such as the mean, variances and covariances. This is called the expanded data set, and, as such, is a concept to be presented and discussed

Objectives

Methods

Results

Conclusion