Enhancing genomic prediction with genome-wide association studies in multiparental maize populations

Abstract

Genome-wide association mapping using dense marker sets has identified some nucleotide variants affecting complex traits that have been validated with fine-mapping and functional analysis. However, many sequence variants associated with complex traits in maize have small effects and low repeatability. In contrast to genome-wide association study (GWAS), genomic prediction (GP) is typically based on models incorporating information from all available markers, rather than modeling effects of individual loci. We considered methods to integrate results of GWASs into GP models in the context of multiple interconnected families. We compared association tests based on a biallelic additive model constraining the effect of a single-nucleotide polymorphism (SNP) to be equal across all families in which it segregates to a model in which the effect of a SNP can vary across families. Association SNPs were then included as fixed effects into a GP model that also included the random effects of the whole genome background. Simulation studies revealed that the effectiveness of this joint approach depends on the extent of polygenicity of the traits. Congruent with this finding, cross-validation studies indicated that GP including the fixed effects of the most significantly associated SNPs along with the polygenic background was more accurate than the polygenic background model alone for moderately complex but not highly polygenic traits measured in the maize nested association mapping population. Individual SNPs with strong and robust association signals can effectively improve GP. Our approach provides a new integrative modeling approach for both reliable gene discovery and robust GP.

A connected half‐sib family training population for genomic prediction in barley

“…In theory, accounting for large effect markers as fixed effects in genomic prediction should improve PA and relative efficiency of selection when heritability is high and the marker explains over 10% of the genetic variance (Bernardo, 2014; Rutkoski et al., 2012). Empirical studies support incorporation of fixed marker effects from diagnostic markers (Daetwyler, Bansal, Bariana, Hayden, & Hayes, 2014) or de novo GWAS (Bian & Holland, 2017; Herter et al., 2019; Moore et al., 2017; Spindel et al., 2016) in the TP to improve PA. In contrast, Rice and Lipka (2018) reported minimal to negative gains in PA from combined genomic prediction and de novo GWAS across the majority of 216 genetic architecture simulations, possibly due to differences in marker effect across TP and VP.…”

“…Prediction accuracy may be improved by including large effect markers as fixed effects in the genomic prediction model (Bernardo, 2014; Rutkoski et al., 2012), including markers discovered through de novo GWAS (Moore et al., 2017; Spindel et al., 2016). Bian and Holland (2017) proposed genomic prediction and GWAS in a nested marker effect model in the original maize NAM to allow for family‐specific allelic effects. Addition of significant marker effects from the nested model increased PA for two complex disease traits, but not for plant height, a highly polygenic trait.…”

Genomic basis of European ash tree resistance to ash dieback fungus

“…The levels of accuracy which our GP reached are high, and comparable to those that are used to inform selections in crop 46–50 , tree 12,51 and livestock breeding programmes 52,53 . Thus, our results have the potential to increase the speed at which we can successfully breed ash dieback resistant trees.…”

Genomewide analysis of biomass responses to water withholding in young plants of maize inbred lines with expired plant variety protection certificate

“…The value of M eff was calculated according to multiple testing method as implemented in SimpleM R script [30,31]. According to the results of Bian and Holland [32] that showed the stable predictive abilities of the loci detected in the range of -log(P) thresholds from 4 to Bonferroni corrected value in oligogenic and polygenic traits, another less-stringent threshold of 4 was applied.…”