Optimising Genomic Selection in Wheat: Effect of Marker Density, Population Size and Population Structure on Prediction Accuracy

Abstract

Genomic selection applied to plant breeding enables earlier estimates of a line’s performance and significant reductions in generation interval. Several factors affecting prediction accuracy should be well understood if breeders are to harness genomic selection to its full potential. We used a panel of 10,375 bread wheat (Triticum aestivum) lines genotyped with 18,101 SNP markers to investigate the effect and interaction of training set size, population structure and marker density on genomic prediction accuracy. Through assessing the effect of training set size we showed the rate at which prediction accuracy increases is slower beyond approximately 2,000 lines. The structure of the panel was assessed via principal component analysis and K-means clustering, and its effect on prediction accuracy was examined through a novel cross-validation analysis according to the K-means clusters and breeding cohorts. Here we showed that accuracy can be improved by increasing the diversity within the training set, particularly when relatedness between training and validation sets is low. The breeding cohort analysis revealed that traits with higher selection pressure (lower allelic diversity) can be more accurately predicted by including several previous cohorts in the training set. The effect of marker density and its interaction with population structure was assessed for marker subsets containing between 100 and 17,181 markers. This analysis showed that response to increased marker density is largest when using a diverse training set to predict between poorly related material. These findings represent a significant resource for plant breeders and contribute to the collective knowledge on the optimal structure of calibration panels for genomic prediction.

Impact of sorghum racial structure and diversity on genomic prediction of grain yield components

“…In this approach, we ran predictions either by dividing individuals from a single race into training and validation folds, or by using individuals from certain race(s) as a training population to predict genetic values of individuals from unrelated race(s) (Figure 1b). Similar strategies have previously been reported for within and across group genomic prediction for diversity panels in maize ( Zea mays L.) and rice ( Oryza sativa L.) (Guo et al., 2014), and for breeding population in wheat ( Triticum aestivum L.) (Norman et al., 2018).…”

“…Cross‐validation approaches similar to the one used in this study have resulted in higher r for within‐population prediction than across‐population prediction in wheat (Norman et al., 2018), rice, and maize (Guo et al., 2014). Although average r across all races in our study was higher for within‐population prediction for most of the traits, the variation in r for individual race and trait combination shows interaction between population structure and trait genetic architecture.…”

“…The implementation of genomic prediction in a diverse and stratified population, however, requires careful consideration of the genomic relationship and population structure (Habier, Fernando, & Dekkers, 2007; Jannink, Lorenz, & Iwata, 2010). Population structure has been shown to affect the accuracy of genomic prediction in stratified populations across several crop species (Guo et al., 2014; Ly et al., 2013; Norman, Taylor, Edwards, & Kuchel, 2018). Population structure analysis can be done using non‐model‐based approaches like principal component analysis (Price et al., 2006) or model‐based clustering approaches like ADMIXTURE (Alexander, Novembre, & Lange, 2009).…”

“…Since interracial diversity is important for heterotic gain, application of genomic prediction in sorghum breeding will benefit from investigations into the effect of racial structure on prediction accuracy. Although the effect of population structure in genomic prediction has been extensively studied in major cereal crops, the distinctive evolutionary history and racial structure of sorghum merits the need for investigation into the effect of sorghum racial structure in genomic prediction (Guo et al., 2014; Isidro et al., 2015; Norman et al., 2018). A previous study examined the effect of genetic relatedness on genomic prediction of pedigreed male lines in a sorghum breeding program; however, there has been no studies on the role of sorghum racial structure in genomic prediction (Hunt, van Eeuwijk, Mace, Hayes, & Jordan, 2018).…”

Genomic Prediction of Pumpkin Hybrid Performance

“…3. From the figure, the prediction accuracy generally appears to improve as the marker density increases from the beginning, which could enhance the chance of marker–QTL associations to capture more accurate marker effects (Desta and Ortiz, 2014; Norman et al, 2018). However, the prediction accuracy gradually reaches the plateau as marker density exceeds 1000, 2000, and 4000 SNPs for the C. maxima intracrossing group, the C. moschata intracrossing group, and their intercrossing group, respectively.…”