Deep Learning for Predicting Complex Traits in Spring Wheat Breeding Program

Abstract

Genomic selection (GS) is transforming the field of plant breeding and implementing models that improve prediction accuracy for complex traits is needed. Analytical methods for complex datasets traditionally used in other disciplines represent an opportunity for improving prediction accuracy in GS. Deep learning (DL) is a branch of machine learning (ML) which focuses on densely connected networks using artificial neural networks for training the models. The objective of this research was to evaluate the potential of DL models in the Washington State University spring wheat breeding program. We compared the performance of two DL algorithms, namely multilayer perceptron (MLP) and convolutional neural network (CNN), with ridge regression best linear unbiased predictor (rrBLUP), a commonly used GS model. The dataset consisted of 650 recombinant inbred lines (RILs) from a spring wheat nested association mapping (NAM) population planted from 2014–2016 growing seasons. We predicted five different quantitative traits with varying genetic architecture using cross-validations (CVs), independent validations, and different sets of SNP markers. Hyperparameters were optimized for DL models by lowering the root mean square in the training set, avoiding model overfitting using dropout and regularization. DL models gave 0 to 5% higher prediction accuracy than rrBLUP model under both cross and independent validations for all five traits used in this study. Furthermore, MLP produces 5% higher prediction accuracy than CNN for grain yield and grain protein content. Altogether, DL approaches obtained better prediction accuracy for each trait, and should be incorporated into a plant breeder’s toolkit for use in large scale breeding programs.

Multi-Trait Machine and Deep Learning Models for Genomic Selection using Spectral Information in a Wheat Breeding Program

“…Grain protein content and grain yield were accurately predicted when spectral data was collected at heading and grain filling stages, respectively (Sandhu et al, 2021b). Similarly, we observed that deep learning based GS models improve prediction accuracy by 3-5% in different agronomic traits in wheat (Sandhu et al, 2021a). The main objectives of this study were to 1) Optimize different MT machine and deep models for predicting grain yield and grain protein content in wheat, 2) Compare the performance of MT-GS and UT-GS models, and 3) Compare the performances of MT-GS mixed models with machine and deep models under cross-validation and independent validation scenarios.…”

“…Deep learning based GS models gave 0-5% higher prediction accuracies for various agronomic traits in wheat in our previous work (Sandhu et al, 2021a). Montesinos-López et al (2018b) concluded that deep learning models were superior for six out of nine traits evaluated in maize and wheat over the traditional GBLUP.…”

“…Initial genotyping data consisted of 73,345 high-quality polymorphic markers anchored to the Chinese Spring RefSeqv1 reference map (Jordan et al, 2018). Detailed procedure about genotyping, SNP calling, map construction, and filtration is described in previous publications (Sandhu et al, 2021a). Quality filtering involved removing monomorphic markers and markers missing more than 20% of the genotyping data.…”

“…The hyperparameters that provided the least MSE on the validation set were selected, and later used on the testing set. Detailed information about the hyperparameter optimization is referred to in a previous publication (Sandhu et al, 2021a). All the MLP algorithms were implemented in Python 3.7 using Keras and Scikit learn libraries (Gulli and Pal, 2017).…”

“…The complete layout for CNN is provided in Figure 1B. Initial values for the hyperparameters were based on previous findings on UT-GS models (Sandhu et al, 2021a). The CNN model used here consists of one input layer, two convolutional layers, two pooling layers, a dense layer, a flatten layer, two dropouts, and an output layer.…”