Genome-wide association mapping studies (GWAS) based on Big Data are a potential approach to improve marker-assisted selection in plant breeding. The number of available phenotypic and genomic data sets in which medium-sized populations of several hundred individuals have been studied is rapidly increasing. Combining these data and using them in GWAS could increase both the power of QTL discovery and the accuracy of estimation of underlying genetic effects, but is hindered by data heterogeneity and lack of interoperability. In this study, we used genomic and phenotypic data sets, focusing on Central European winter wheat populations evaluated for heading date. We explored strategies for integrating these data and subsequently the resulting potential for GWAS. Establishing interoperability between data sets was greatly aided by some overlapping genotypes and a linear relationship between the different phenotyping protocols, resulting in high quality integrated phenotypic data. In this context, genomic prediction proved to be a suitable tool to study relevance of interactions between genotypes and experimental series, which was low in our case. Contrary to expectations, fewer associations between markers and traits were found in the larger combined data than in the individual experimental series. However, the predictive power based on the marker-trait associations of the integrated data set was higher across data sets. Therefore, the results show that the integration of medium-sized to Big Data is an approach to increase the power to detect QTL in GWAS. The results encourage further efforts to standardize and share data in the plant breeding community.

To extend the contemporary understanding into the grain yield heterosis of wheat, the current study investigated the contribution of deleterious alleles in shaping mid-parent heterosis (MPH). These alleles occur at low frequency in the genome and are often missed by automated genotyping platforms like SNP arrays. The deleterious alleles herein were detected using a quantitative measurement of evolutionary conservation based on the phylogeny of wheat and investigations were made to: (1) assess the benefit of including deleterious alleles into MPH prediction models and (2) understand the genetic underpinnings of deleterious SNPs for grain yield MPH using contrasting crosses viz. elite × elite (Exp. 1) and elite × plant genetic resources (PGR; Exp. 2). In our study, we found a lower allele frequency of moderately deleterious alleles in elites compared to PGRs. This highlights the role of purifying selection for the development of elite wheat cultivars. It was shown that deleterious alleles are informative for MPH prediction models: modelling their additive-by-additive effects in Exp. 1 and dominance as well as associated digenic epistatic effects in Exp. 2 significantly boosts prediction accuracies of MPH. Furthermore, heterotic- quantitative trait loci's underlying MPH was investigated and their properties were contrasted in the two crosses. Conclusively, it was proposed that incomplete dominance of deleterious alleles contributes to grain yield heterosis in elite crosses (Exp. 1).

Extensive exotic introgression could significantly enlarge the genetic distance of hybrid parental populations to promote strong heterosis. The goal of this study was to investigate whether genome-wide prediction can support pre-breeding in populations with exotic introgressions. We evaluated seed yield, seed yield related traits and seed quality traits of 363 hybrids of Brassica napus (AACC) derived from two parental populations divergent on massive exotic introgression of related species in three environments. The hybrids presented strong heterosis on seed yield, which was much higher than other investigated traits. Five genomic best linear unbiased prediction models considering the exotic introgression and different marker effects (additive, dominance, and epistatic effects) were constructed to test the prediction ability for different traits of the hybrids. The analysis showed that the trait complexity, exotic introgression, genetic relationship between the training set and testing set, training set size, and environments affected the prediction ability. The models with best prediction ability for different traits varied. However, relatively high prediction ability (e.g., 0.728 for seed yield) was also observed when the simplest models were used, excluding the effects of the special exotic introgression and epistasis effect by 5-fold cross validation, which would simplify the prediction for the trait with complex architecture for hybrids with exotic introgression. The results provide novel insights and strategies for genome-wide prediction of hybrids between genetically distinct parent groups with exotic introgressions.

Genome-wide prediction is a promising approach to boost selection gain in hybrid breeding. Our main objective was to evaluate the potential and limits of genome-wide prediction to identify superior hybrid combinations adapted to Northwest China. A total of 490 hybrids derived from crosses among 119 inbred lines from the Shaan A and Shaan B heterotic pattern were used for genome-wide prediction of ten agronomic traits. We tested eight different statistical prediction models considering additive (A) effects and in addition evaluated the impact of dominance (D) and epistasis (E) on the prediction ability. Employing five-fold cross validation, we show that the average prediction ability ranged from 0.386 to 0.794 across traits and models. Six parametric methods, i.e. ridge regression, LASSO, Elastic Net, Bayes B, Bayes C and reproducing kernel Hilbert space (RKHS) approach, displayed a very similar prediction ability for each trait and two non-parametric methods (random forest and support vector machine) had a higher prediction performance for the trait rind penetrometer resistance of the third internode above ground (RPR_TIAG). The models of A + D RKHS and A + D + E RKHS were slightly better for predicting traits with a relatively high non-additive variance. Integrating trait-specific markers into the A + D RKHS model improved the prediction ability of grain yield by 3%, from 0.528 to 0.558. Of all 6328 potential hybrids, selection of the top 44 hybrids would lead to a 6% increase in grain yield compared with Zhengdan 958, a commercially successful hybrid variety. In conclusion, our results substantiate the value of genome-wide prediction for hybrid breeding and suggest dozens of promising single crosses for developing high-yielding hybrids for Northwest China.