AUTHOR=Montesinos-López Osval A. , Montesinos-López Abelardo , Tuberosa Roberto , Maccaferri Marco , Sciara Giuseppe , Ammar Karim , Crossa José 

TITLE=Multi-Trait, Multi-Environment Genomic Prediction of Durum Wheat With Genomic Best Linear Unbiased Predictor and Deep Learning Methods

JOURNAL=Frontiers in Plant Science

VOLUME=Volume 10 - 2019

YEAR=2019

URL=https://www.frontiersin.org/journals/plant-science/articles/10.3389/fpls.2019.01311

DOI=10.3389/fpls.2019.01311

ISSN=1664-462X

ABSTRACT=Although durum wheat (Triticum turgidum var. durum Desf.)   is a minor cereal crop representing just 5% of the total wheat crop, it is a staple food for Mediterranean people who use it to make pasta, couscous, bulgur and bread. For this reason, there are research programs (mostly in Europe) working to improve durum wheat production and quality. In this paper we cover multi-trait prediction of grain yield (GY), days to heading (DH) and plant height (PH) of 270 durum wheat lines that were evaluated in 43 environments (location-year combinations) in Bologna, Italy. Multi-trait prediction analyses were performed by implementing a multi-trait deep learning model with a feed-forward network architecture and a rectified linear unit activation function with a grid search approach for the selection of hyper-parameters. The results of the multi-trait deep learning method also were compared with univariate predictions of the genomic best linear unbiased predictor (GBLUP) method and the univariate counterpart of the multi-trait deep learning method. All models were implemented with and without the genotype×environment interaction term. We found that the best predictions were observed without the genotype×environment interaction term in the univariate and multivariate deep learning methods, but under the GBLUP method, the best predictions were observed taking into account the interaction term. We also found that in general the best predictions were observed under the GBLUP model but the predictions of the multi-trait deep learning model were very similar to those of the GBLUP model. This result provides more evidence that the GBLUP model is a powerful approach for genomic prediction, but also that the deep learning method is a practical approach for predicting univariate and multivariate traits in the context of genomic selection.