<p>With the rapid development of sequencing technology, the application of genomic prediction has become more and more common in breeding schemes of livestocks and crops. Selecting an appropriate statistical model is of central importance to achieve high prediction accuracy. Recently, machine learning models have been expected to upgrade genomic prediction into a new era. However, the perspective still suffers from lack of evidence that machine learning models can generally outperform the traditional ones on empirical data sets. In this study, we compared two machine learning models based on artificial neural network (ANN) and convolutional neural network (CNN) with four traditional models, including genomic best linear unbiased prediction (GBLUP), Bayesian ridge regression (BRR), BayesA and BayesB, using three published data sets for grain yield in wheat. For each model, we considered two variants: modeling and ignoring the genotype-by-environment (<InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11032_2025_1598_Article_IEq1.gif" Format="GIF" Height="15" Rendition="HTML" Resolution="72" Type="Linedraw" Width="51" /> </InlineMediaObject> <EquationSource Format="TEX">\(G \times E\)</EquationSource> </InlineEquation>) interaction. In the comparison, we considered two strategies of cross-validation: predicting genotypes that have not been evaluated in any environment (CV1) and predicting genotypes that have been tested in other environments (CV2). Our results showed that traditional Bayesian models (BayesA, BayesB, and BRR) outperformed GBLUP, ANN and CNN when considering <InlineEquation ID="IEq2"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11032_2025_1598_Article_IEq1.gif" Format="GIF" Height="15" Rendition="HTML" Resolution="72" Type="Linedraw" Width="51" /> </InlineMediaObject> <EquationSource Format="TEX">\(G \times E\)</EquationSource> </InlineEquation> interaction. The accuracies of ANN and CNN were higher than traditional models only in CV1 and when <InlineEquation ID="IEq3"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11032_2025_1598_Article_IEq1.gif" Format="GIF" Height="15" Rendition="HTML" Resolution="72" Type="Linedraw" Width="51" /> </InlineMediaObject> <EquationSource Format="TEX">\(G \times E\)</EquationSource> </InlineEquation> interaction was ignored. It was also found that the performance of the two machine learning models was significantly affected by the interaction between the CV strategy and the way of treating the <InlineEquation ID="IEq4"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11032_2025_1598_Article_IEq1.gif" Format="GIF" Height="15" Rendition="HTML" Resolution="72" Type="Linedraw" Width="51" /> </InlineMediaObject> <EquationSource Format="TEX">\(G \times E\)</EquationSource> </InlineEquation> interaction, while that of the four traditional models was only influenced by whether the <InlineEquation ID="IEq5"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="11032_2025_1598_Article_IEq1.gif" Format="GIF" Height="15" Rendition="HTML" Resolution="72" Type="Linedraw" Width="51" /> </InlineMediaObject> <EquationSource Format="TEX">\(G \times E\)</EquationSource> </InlineEquation> interaction was considered or not. Thus, machine learning models can be a powerful complementary to the traditional ones and their superiority may depend on the prediction scenario. Among the two machine learning models, we observed that the accuracy of ANN was higher than CNN in most cases, indicating that it is still challenging to adapt complex machine learning models such as CNN to genomic prediction.</p>

Comparing artificial and convolutional neural networks with traditional models for Genomic prediction in wheat

期刊论文
发表日期：

2025年9月10日

查看全文

Wei Zhao,
Jie Sheng

摘要

With the rapid development of sequencing technology, the application of genomic prediction has become more and more common in breeding schemes of livestocks and crops. Selecting an appropriate statistical model is of central importance to achieve high prediction accuracy. Recently, machine learning models have been expected to upgrade genomic prediction into a new era. However, the perspective still suffers from lack of evidence that machine learning models can generally outperform the traditional ones on empirical data sets. In this study, we compared two machine learning models based on artificial neural network (ANN) and convolutional neural network (CNN) with four traditional models, including genomic best linear unbiased prediction (GBLUP), Bayesian ridge regression (BRR), BayesA and BayesB, using three published data sets for grain yield in wheat. For each model, we considered two variants: modeling and ignoring the genotype-by-environment ( \(G \times E\) ) interaction. In the comparison, we considered two strategies of cross-validation: predicting genotypes that have not been evaluated in any environment (CV1) and predicting genotypes that have been tested in other environments (CV2). Our results showed that traditional Bayesian models (BayesA, BayesB, and BRR) outperformed GBLUP, ANN and CNN when considering \(G \times E\) interaction. The accuracies of ANN and CNN were higher than traditional models only in CV1 and when \(G \times E\) interaction was ignored. It was also found that the performance of the two machine learning models was significantly affected by the interaction between the CV strategy and the way of treating the \(G \times E\) interaction, while that of the four traditional models was only influenced by whether the \(G \times E\) interaction was considered or not. Thus, machine learning models can be a powerful complementary to the traditional ones and their superiority may depend on the prediction scenario. Among the two machine learning models, we observed that the accuracy of ANN was higher than CNN in most cases, indicating that it is still challenging to adapt complex machine learning models such as CNN to genomic prediction.