Genome‐enabled prediction through machine learning methods considering different levels of trait complexity

Abstract

Genomic‐wide selection (GWS) consists of the use of a large number of molecular markers for the prediction of genetic values and has been shown to be highly relevant for genetic improvement. The objective of this work was to evaluate and compare the predictive performance of statistical (ridge regression‐best linear unbiased predictor [RR‐BLUP] and BayesB) and machine learning methods through GWS in simulated populations with traits presenting different levels of heritability and quantitative trait loci (QTL) numbers in the presence of dominant and epistatic effects. The simulated genome of population F2 was formed by 1,000 individuals and genotyped with 2,010 single nucleotide polymorphism (SNP) markers. Twenty‐six traits were simulated considering QTL numbers ranging from two to 88 and heritabilities of .3 and .6. The selective and predictive performances were evaluated using the multilayer perceptron (MLP), radial basis function (RBF), decision trees (DT), bagging (BA), random forest (RF), and boosting (BO) machine learning models and the classical RR‐BLUP and BayesB methods. A high effect of heritability was observed for the results of selective accuracy when compared to the increased QTL number. In addition, the selective accuracy based on the number of QTL demonstrates that the application of alternative machine learning models, such as RBF, BA, BO, and RF, can be suitable for the analysis according to QTL number. Machine learning methods are powerful tools for predicting genetic values with epistatic gene control in traits with different degrees of heritability and different numbers of controlling genes.

Prediction of the importance of auxiliary traits using computational intelligence and machine learning: A simulation study

“…Machine learning algorithms have the advantage of modeling data in a non-linear and non-parametric manner [ 60 ]. Unlike many traditional statistical methods, these algorithms are built with the advantage of dealing with noisy, complex, and heterogeneous data [ 61 – 64 ] reported that machine learning methods are powerful tools for predicting genetic values with epistatic genetic control in traits with different degrees of heritability and different numbers of controlling genes. The results obtained in the present study can be used to select genotypes and test them in the field.…”

Low‐coverage whole‐genome sequencing reveals molecular markers for spawning season and sex identification in Gulf of Maine Atlantic cod (Gadus morhua, Linnaeus 1758)

“…outperformed Bayesian or linear mixed model techniques when relatively few qualitative trait loci (2-8) were simulated, but at increased trait complexities, all techniques were equally predictive. Other factors beyond trait complexity like heritance and dominance seem to play a significant role in determining the performance of machine learning approaches when identifying SNP loci (Alves et al, 2020;Barbosa et al, 2021).…”

“…Conversely, sex determination in Atlantic cod is controlled by a single gene (Kirubakaran et al, 2019), making it a far simpler trait. Machine learning techniques like GRRF may be well-suited to selecting loci associated with complex traits because all loci, and complex interactions between them are included in the adjustment of the model, allowing for the contribution of unassociated, but highly correlated loci (Barbosa et al, 2021). Simulated and empirical research has shown that incorporating machine learning methods (including random forest) into genome prediction and identifying relevant genomic regions of complex traits improves accuracy relative to linear mixed models and Bayesian approaches (Maldonado et al, 2020;Yin et al, 2020).…”

Change Is Upon Us, and We Need to Adapt

“…It is exciting to see the first uses of deep learning tools for stress analysis (Nagasubramanian et al., 2022), computer vision for maize ear phenotyping (Gonzalez et al., 2022), and image‐based oat panicle phenotyping (Berro et al., 2022). The advances in genomic prediction have also only been possible with new machine‐learning tools (Barbosa et al., 2021), and they are now being powerfully used to characterize genomic diversity measures using more data (da Costa et al., 2022).…”