Integrating Crop Growth Models with Whole Genome Prediction through Approximate Bayesian Computation

Abstract

Genomic selection, enabled by whole genome prediction (WGP) methods, is revolutionizing plant breeding. Existing WGP methods have been shown to deliver accurate predictions in the most common settings, such as prediction of across environment performance for traits with additive gene effects. However, prediction of traits with non-additive gene effects and prediction of genotype by environment interaction (G×E), continues to be challenging. Previous attempts to increase prediction accuracy for these particularly difficult tasks employed prediction methods that are purely statistical in nature. Augmenting the statistical methods with biological knowledge has been largely overlooked thus far. Crop growth models (CGMs) attempt to represent the impact of functional relationships between plant physiology and the environment in the formation of yield and similar output traits of interest. Thus, they can explain the impact of G×E and certain types of non-additive gene effects on the expressed phenotype. Approximate Bayesian computation (ABC), a novel and powerful computational procedure, allows the incorporation of CGMs directly into the estimation of whole genome marker effects in WGP. Here we provide a proof of concept study for this novel approach and demonstrate its use with synthetic data sets. We show that this novel approach can be considerably more accurate than the benchmark WGP method GBLUP in predicting performance in environments represented in the estimation set as well as in previously unobserved environments for traits determined by non-additive gene effects. We conclude that this proof of concept demonstrates that using ABC for incorporating biological knowledge in the form of CGMs into WGP is a very promising and novel approach to improving prediction accuracy for some of the most challenging scenarios in plant breeding and applied genetics.

Use of Crop Growth Models with Whole‐Genome Prediction: Application to a Maize Multienvironment Trial

“…This extension of WGP was achieved within a high performance computing environment by application of an approximate Bayesian computation (ABC; Marjoram et al, 2014) algorithm that used a suitable CGM in place of the likelihood function to obtain the WGP outcomes; we identify the WGP methodology with the CGM embedded within the prediction algorithm as CGM-WGP. The simulation results reported by Technow et al (2015) were encouraging. It is necessary to consider how to scale the proposed methods to the empirical data sets generated by breeding programs.…”

“…If this can be achieved, the breeder could consider predicted performance of genotypes for key environment types of the TPE and move beyond predictions based solely on marginal performance across environments. Technow et al (2015) used simulation to demonstrate how CGMs could be applied with WGP for such purposes. This extension of WGP was achieved within a high performance computing environment by application of an approximate Bayesian computation (ABC; Marjoram et al, 2014) algorithm that used a suitable CGM in place of the likelihood function to obtain the WGP outcomes; we identify the WGP methodology with the CGM embedded within the prediction algorithm as CGM-WGP.…”

“…It is necessary to consider how to scale the proposed methods to the empirical data sets generated by breeding programs. This paper complements the proof-of-concept paper of Technow et al (2015) by applying the CGM-WGP methodology to an empirical maize data set and focuses on key practical considerations that are involved in the reduction-to-practice of the CGM-WGP methodology.…”

Introduction to a Special Issue on Genotype by Environment Interaction

“…Advances in the identification of loci affecting quantitative traits (QTL) and in the use of crop growth models (CGM) lead us to posit the answer is "yes." The applied breeder now has unprecedented ability to manipulate genes identified as mechanistically involved in G ´ E. Likewise, interactions that are understood at morphological and physiological levels can be predicted using CGM, leading to target ideotypes defined phenotypically, as promoted for many years (Donald, 1968), or genetically as in new approaches (Technow et al, 2015).…”

Crop Models as Research and Interpretative Tools

“…Only modeling allows for effectively analyzing the full spectrum of the G ½ M ½ E combinations and provides a rational basis for development and testing of novel wheat ideotypes optimized for target agricultural landscapes and future climatic conditions. The next generation models should comprise the so-called genetic coefficients for modeling differences between hybrids (genebased crop model) [38]. Genomic forecasting using agricultural crop models is capable of performing better than statistical methods using only genetic data [39].…”