Importance of genetic architecture in marker selection decisions for genomic prediction.

KEY MESSAGE: We demonstrate potential for improved multi-environment genomic prediction accuracy using structural variant markers. However, the degree of observed improvement is highly dependent on the genetic architecture of the trait. Breeders commonly use genetic markers to predict the performance of untested individuals as a way to improve the efficiency of breeding programs. These genomic prediction models have almost exclusively used single nucleotide polymorphisms (SNPs) as their source of genetic information, even though other types of markers exist, such as structural variants (SVs). Given that SVs are associated with environmental adaptation and not all of them are in linkage disequilibrium to SNPs, SVs have the potential to bring additional information to multi-environment prediction models that are not captured by SNPs alone. Here, we evaluated different marker types (SNPs and/or SVs) on prediction accuracy across a range of genetic architectures for simulated traits across multiple environments. Our results show that SVs can improve prediction accuracy, but it is highly dependent on the genetic architecture of the trait and the relative gain in accuracy is minimal. When SVs are the only causative variant type, 70% of the time SV predictors outperform SNP predictors. However, the improvement in accuracy in these instances is only 1.5% on average. Further simulations with predictors in varying degrees of LD with causative variants of different types (e.g., SNPs, SVs, SNPs and SVs) showed that prediction accuracy increased as linkage disequilibrium between causative variants and predictors increased regardless of the marker type. This study demonstrates that knowing the genetic architecture of a trait in deciding what markers to use in large-scale genomic prediction modeling in a breeding program is more important than what types of markers to use.

Linking genetic and environmental factors through marker effect networks to understand trait plasticity.

Understanding how plants adapt to specific environmental changes and identifying genetic markers associated with phenotypic plasticity can help breeders develop plant varieties adapted to a rapidly changing climate. Here, we propose the use of marker effect networks as a novel method to identify markers associated with environmental adaptability. These marker effect networks are built by adapting commonly used software for building gene coexpression networks with marker effects across growth environments as the input data into the networks. To demonstrate the utility of these networks, we built networks from the marker effects of ∼2,000 nonredundant markers from 400 maize hybrids across 9 environments. We demonstrate that networks can be generated using this approach, and that the markers that are covarying are rarely in linkage disequilibrium, thus representing higher biological relevance. Multiple covarying marker modules associated with different weather factors throughout the growing season were identified within the marker effect networks. Finally, a factorial test of analysis parameters demonstrated that marker effect networks are relatively robust to these options, with high overlap in modules associated with the same weather factors across analysis parameters. This novel application of network analysis provides unique insights into phenotypic plasticity and specific environmental factors that modulate the genome.

Assessment of two statistical approaches for variance genome-wide association studies in plants.

Genomic loci that control the variance of agronomically important traits are increasingly important due to the profusion of unpredictable environments arising from climate change. The ability to identify such variance-controlling loci in association studies will be critical for future breeding efforts. Two statistical approaches that have already been used in the variance genome-wide association study (vGWAS) paradigm are the Brown-Forsythe test (BFT) and the double generalized linear model (DGLM). To ensure that these approaches are deployed as effectively as possible, it is critical to study the factors that influence their ability to identify variance-controlling loci. We used genome-wide marker data in maize (Zea mays L.) and Arabidopsis thaliana to simulate traits controlled by epistasis, genotype by environment (GxE) interactions, and variance quantitative trait nucleotides (vQTNs). We then quantified true and false positive detection rates of the BFT and DGLM across all simulated traits. We also conducted a vGWAS using both the BFT and DGLM on plant height in a maize diversity panel. The observed true positive detection rates at the maximum sample size considered (N = 2815) suggest that both of these vGWAS approaches are capable of identifying epistasis and GxE for sufficiently large sample sizes. We also noted that the DGLM decisively outperformed the BFT for simulated traits controlled by vQTNs at sample sizes of N = 500. Although we conclude that there are still certain aspects of vGWAS approaches that need further refinement, this study suggests that the BFT and DGLM are capable of identifying variance-controlling loci in current state-of-the-art plant or agronomic data sets.

A multi-trait multi-locus stepwise approach for conducting GWAS on correlated traits.

The ability to accurately quantify the simultaneous effect of multiple genomic loci on multiple traits is now possible due to current and emerging high-throughput genotyping and phenotyping technologies. To date, most efforts to quantify these genotype-to-phenotype relationships have focused on either multi-trait models that test a single marker at a time or multi-locus models that quantify associations with a single trait. Therefore, the purpose of this study was to compare the performance of a multi-trait, multi-locus stepwise (MSTEP) model selection procedure we developed to (a) a commonly used multi-trait single-locus model and (b) a univariate multi-locus model. We used real marker data in maize (Zea mays L.) and soybean (Glycine max L.) to simulate multiple traits controlled by various combinations of pleiotropic and nonpleiotropic quantitative trait nucleotides (QTNs). In general, we found that both multi-trait models outperformed the univariate multi-locus model, especially when analyzing a trait of low heritability. For traits controlled by either a combination of pleiotropic and nonpleiotropic QTNs or a large number of QTNs (i.e., 50), our MSTEP model often outperformed at least one of the two alternative models. When applied to the analysis of two tocochromanol-related traits in maize grain, MSTEP identified the same peak-associated marker that has been reported in a previous study. We therefore conclude that MSTEP is a useful addition to the suite of statistical models that are commonly used to gain insight into the genetic architecture of agronomically important traits.

Optical topometry and machine learning to rapidly phenotype stomatal patterning traits for maize QTL mapping.

Stomata are adjustable pores on leaf surfaces that regulate the tradeoff of CO2 uptake with water vapor loss, thus having critical roles in controlling photosynthetic carbon gain and plant water use. The lack of easy, rapid methods for phenotyping epidermal cell traits have limited discoveries about the genetic basis of stomatal patterning. A high-throughput epidermal cell phenotyping pipeline is presented here and used for quantitative trait loci (QTL) mapping in field-grown maize (Zea mays). The locations and sizes of stomatal complexes and pavement cells on images acquired by an optical topometer from mature leaves were automatically determined. Computer estimated stomatal complex density (SCD; R2 = 0.97) and stomatal complex area (SCA; R2 = 0.71) were strongly correlated with human measurements. Leaf gas exchange traits were genetically correlated with the dimensions and proportions of stomatal complexes (rg = 0.39-0.71) but did not correlate with SCD. Heritability of epidermal traits was moderate to high (h2 = 0.42-0.82) across two field seasons. Thirty-six QTL were consistently identified for a given trait in both years. Twenty-four clusters of overlapping QTL for multiple traits were identified, with univariate versus multivariate single marker analysis providing evidence consistent with pleiotropy in multiple cases. Putative orthologs of genes known to regulate stomatal patterning in Arabidopsis (Arabidopsis thaliana) were located within some, but not all, of these regions. This study demonstrates how discovery of the genetic basis for stomatal patterning can be accelerated in maize, a C4 model species where these processes are poorly understood.

Machine learning-enabled phenotyping for GWAS and TWAS of WUE traits in 869 field-grown sorghum accessions.

Sorghum (Sorghum bicolor) is a model C4 crop made experimentally tractable by extensive genomic and genetic resources. Biomass sorghum is studied as a feedstock for biofuel and forage. Mechanistic modeling suggests that reducing stomatal conductance (gs) could improve sorghum intrinsic water use efficiency (iWUE) and biomass production. Phenotyping to discover genotype-to-phenotype associations remains a bottleneck in understanding the mechanistic basis for natural variation in gs and iWUE. This study addressed multiple methodological limitations. Optical tomography and a machine learning tool were combined to measure stomatal density (SD). This was combined with rapid measurements of leaf photosynthetic gas exchange and specific leaf area (SLA). These traits were the subject of genome-wide association study and transcriptome-wide association study across 869 field-grown biomass sorghum accessions. The ratio of intracellular to ambient CO2 was genetically correlated with SD, SLA, gs, and biomass production. Plasticity in SD and SLA was interrelated with each other and with productivity across wet and dry growing seasons. Moderate-to-high heritability of traits studied across the large mapping population validated associations between DNA sequence variation or RNA transcript abundance and trait variation. A total of 394 unique genes underpinning variation in WUE-related traits are described with higher confidence because they were identified in multiple independent tests. This list was enriched in genes whose Arabidopsis (Arabidopsis thaliana) putative orthologs have functions related to stomatal or leaf development and leaf gas exchange, as well as genes with nonsynonymous/missense variants. These advances in methodology and knowledge will facilitate improving C4 crop WUE.

Phenotyping stomatal closure by thermal imaging for GWAS and TWAS of water use efficiency-related genes.

Stomata allow CO2 uptake by leaves for photosynthetic assimilation at the cost of water vapor loss to the atmosphere. The opening and closing of stomata in response to fluctuations in light intensity regulate CO2 and water fluxes and are essential for maintaining water-use efficiency (WUE). However, a little is known about the genetic basis for natural variation in stomatal movement, especially in C4 crops. This is partly because the stomatal response to a change in light intensity is difficult to measure at the scale required for association studies. Here, we used high-throughput thermal imaging to bypass the phenotyping bottleneck and assess 10 traits describing stomatal conductance (gs) before, during and after a stepwise decrease in light intensity for a diversity panel of 659 sorghum (Sorghum bicolor) accessions. Results from thermal imaging significantly correlated with photosynthetic gas exchange measurements. gs traits varied substantially across the population and were moderately heritable (h2 up to 0.72). An integrated genome-wide and transcriptome-wide association study identified candidate genes putatively driving variation in stomatal conductance traits. Of the 239 unique candidate genes identified with the greatest confidence, 77 were putative orthologs of Arabidopsis (Arabidopsis thaliana) genes related to functions implicated in WUE, including stomatal opening/closing (24 genes), stomatal/epidermal cell development (35 genes), leaf/vasculature development (12 genes), or chlorophyll metabolism/photosynthesis (8 genes). These findings demonstrate an approach to finding genotype-to-phenotype relationships for a challenging trait as well as candidate genes for further investigation of the genetic basis of WUE in a model C4 grass for bioenergy, food, and forage production.

Can improved canopy light transmission ameliorate loss of photosynthetic efficiency in the shade? An investigation of natural variation in Sorghum bicolor.

Previous studies have found that maximum quantum yield of CO2 assimilation (Φ CO2,max,app) declines in lower canopies of maize and miscanthus, a maladaptive response to self-shading. These observations were limited to single genotypes, leaving it unclear whether the maladaptive shade response is a general property of this C4 grass tribe, the Andropogoneae. We explored the generality of this maladaptation by testing the hypothesis that erect leaf forms (erectophiles), which allow more light into the lower canopy, suffer less of a decline in photosynthetic efficiency than drooping leaf (planophile) forms. On average, Φ CO2,max,app declined 27% in lower canopy leaves across 35 accessions, but the decline was over twice as great in planophiles than in erectophiles. The loss of photosynthetic efficiency involved a decoupling between electron transport and assimilation. This was not associated with increased bundle sheath leakage, based on 13C measurements. In both planophiles and erectophiles, shaded leaves had greater leaf absorptivity and lower activities of key C4 enzymes than sun leaves. The erectophile form is considered more productive because it allows a more effective distribution of light through the canopy to support photosynthesis. We show that in sorghum, it provides a second benefit, maintenance of higher Φ CO2,max,app to support efficient use of that light resource.

Comparative evolutionary genetics of deleterious load in sorghum and maize.

Sorghum and maize share a close evolutionary history that can be explored through comparative genomics1,2. To perform a large-scale comparison of the genomic variation between these two species, we analysed ~13 million variants identified from whole-genome resequencing of 499 sorghum lines together with 25 million variants previously identified in 1,218 maize lines. Deleterious mutations in both species were prevalent in pericentromeric regions, enriched in non-syntenic genes and present at low allele frequencies. A comparison of deleterious burden between sorghum and maize revealed that sorghum, in contrast to maize, departed from the domestication-cost hypothesis that predicts a higher deleterious burden among domesticates compared with wild lines. Additionally, sorghum and maize population genetic summary statistics were used to predict a gene deleterious index with an accuracy greater than 0.5. This research represents a key step towards understanding the evolutionary dynamics of deleterious variants in sorghum and provides a comparative genomics framework to start prioritizing these variants for removal through genome editing and breeding.

simplePHENOTYPES: SIMulation of pleiotropic, linked and epistatic phenotypes.

BACKGROUND: Advances in genotyping and phenotyping techniques have enabled the acquisition of a great amount of data. Consequently, there is an interest in multivariate statistical analyses that identify genomic regions likely to contain causal mutations affecting multiple traits (i.e., pleiotropy). As the demand for multivariate analyses increases, it is imperative that optimal tools are available to assess their performance. To facilitate the testing and validation of these multivariate approaches, we developed simplePHENOTYPES, an R/CRAN package that simulates pleiotropy, partial pleiotropy, and spurious pleiotropy in a wide range of genetic architectures, including additive, dominance and epistatic models. RESULTS: We illustrate simplePHENOTYPES' ability to simulate thousands of phenotypes in less than one minute. We then provide two vignettes illustrating how to simulate sets of correlated traits in simplePHENOTYPES. Finally, we demonstrate the use of results from simplePHENOTYPES in a standard GWAS software, as well as the equivalence of simulated phenotypes from simplePHENOTYPES and other packages with similar capabilities. CONCLUSIONS: simplePHENOTYPES is a R/CRAN package that makes it possible to simulate multiple traits controlled by loci with varying degrees of pleiotropy. Its ability to interface with both commonly-used marker data formats and downstream quantitative genetics software and packages should facilitate a rigorous assessment of both existing and emerging statistical GWAS and GS approaches. simplePHENOTYPES is also available at https://github.com/samuelbfernandes/simplePHENOTYPES .

Genetic variation associated with PPO-inhibiting herbicide tolerance in sorghum.

Herbicide application is crucial for weed management in most crop production systems, but for sorghum herbicide options are limited. Sorghum is sensitive to residual protoporphyrinogen oxidase (PPO)-inhibiting herbicides, such as fomesafen, and a long re-entry period is required before sorghum can be planted after its application. Improving sorghum for tolerance to such residual herbicides would allow for increased sorghum production and the expansion of herbicide options for growers. In this study, we observed sorghum tolerance to residual fomesafen. To investigate the underlying tolerance mechanism a genome-wide association mapping study was conducted using field-collected sorghum biomass panel (SBP) data, and a greenhouse assay was developed to confirm the field phenotypes. A total of 26 significant SNPs (FDR<0.05), spanning a 215.3 kb region on chromosome 3, were detected. The ten most significant SNPs included two in genic regions (Sobic.003G136800, and Sobic.003G136900) and eight SNPs in the intergenic region encompassing the genes Sobic.003G136700, Sobic.003G136800, Sobic.003G137000, Sobic.003G136900, and Sobic.003G137100. The gene Sobic.003G137100 (PPXI), which encodes the PPO1 enzyme, one of the targets of PPO-inhibiting herbicides, was located 12kb downstream of the significant SNP S03_13152838. We found that PPXI is highly conserved in sorghum and expression does not significantly differ between tolerant and sensitive sorghum lines. Our results suggest that PPXI most likely does not underlie the observed herbicide tolerance. Instead, the mechanism underlying herbicide tolerance in the SBP is likely metabolism-based resistance, possibly regulated by the action of multiple genes. Further research is necessary to confirm candidate genes and their functions.

Multi-Trait Genome-Wide Association Studies Reveal Loci Associated with Maize Inflorescence and Leaf Architecture.

Maize inflorescence is a complex phenotype that involves the physical and developmental interplay of multiple traits. Given the evidence that genes could pleiotropically contribute to several of these traits, we used publicly available maize data to assess the ability of multivariate genome-wide association study (GWAS) approaches to identify pleiotropic quantitative trait loci (pQTL). Our analysis of 23 publicly available inflorescence and leaf-related traits in a diversity panel of n = 281 maize lines genotyped with 376,336 markers revealed that the two multivariate GWAS approaches we tested were capable of identifying pQTL in genomic regions coinciding with similar associations found in previous studies. We then conducted a parallel simulation study on the same individuals, where it was shown that multivariate GWAS approaches yielded a higher true-positive quantitative trait nucleotide (QTN) detection rate than comparable univariate approaches for all evaluated simulation settings except for when the correlated simulated traits had a heritability of 0.9. We therefore conclude that the implementation of state-of-the-art multivariate GWAS approaches is a useful tool for dissecting pleiotropy and their more widespread implementation could facilitate the discovery of genes and other biological mechanisms underlying maize inflorescence.

Conserved defense responses between maize and sorghum to Exserohilum turcicum.

BACKGROUND: Exserohilum turcicum is an important pathogen of both sorghum and maize, causing sorghum leaf blight and northern corn leaf blight. Because the same pathogen can infect and cause major losses for two of the most important grain crops, it is an ideal pathosystem to study plant-pathogen evolution and investigate shared resistance mechanisms between the two plant species. To identify sorghum genes involved in the E. turcicum response, we conducted a genome-wide association study (GWAS). RESULTS: Using the sorghum conversion panel evaluated across three environments, we identified a total of 216 significant markers. Based on physical linkage with the significant markers, we detected a total of 113 unique candidate genes, some with known roles in plant defense. Also, we compared maize genes known to play a role in resistance to E. turcicum with the association mapping results and found evidence of genes conferring resistance in both crops, providing evidence of shared resistance between maize and sorghum. CONCLUSIONS: Using a genetics approach, we identified shared genetic regions conferring resistance to E. turcicum in both maize and sorghum. We identified several promising candidate genes for resistance to leaf blight in sorghum, including genes related to R-gene mediated resistance. We present significant advancements in the understanding of host resistance to E. turcicum, which is crucial to reduce losses due to this important pathogen.

How Well Can Multivariate and Univariate GWAS Distinguish Between True and Spurious Pleiotropy?

Quantification of the simultaneous contributions of loci to multiple traits, a phenomenon called pleiotropy, is facilitated by the increased availability of high-throughput genotypic and phenotypic data. To understand the prevalence and nature of pleiotropy, the ability of multivariate and univariate genome-wide association study (GWAS) models to distinguish between pleiotropic and non-pleiotropic loci in linkage disequilibrium (LD) first needs to be evaluated. Therefore, we used publicly available maize and soybean genotypic data to simulate multiple pairs of traits that were either (i) controlled by quantitative trait nucleotides (QTNs) on separate chromosomes, (ii) controlled by QTNs in various degrees of LD with each other, or (iii) controlled by a single pleiotropic QTN. We showed that multivariate GWAS could not distinguish between QTNs in LD and a single pleiotropic QTN. In contrast, a unique QTN detection rate pattern was observed for univariate GWAS whenever the simulated QTNs were in high LD or pleiotropic. Collectively, these results suggest that multivariate and univariate GWAS should both be used to infer whether or not causal mutations underlying peak GWAS associations are pleiotropic. Therefore, we recommend that future studies use a combination of multivariate and univariate GWAS models, as both models could be useful for identifying and narrowing down candidate loci with potential pleiotropic effects for downstream biological experiments.

Novel Bayesian Networks for Genomic Prediction of Developmental Traits in Biomass Sorghum.

The ability to connect genetic information between traits over time allow Bayesian networks to offer a powerful probabilistic framework to construct genomic prediction models. In this study, we phenotyped a diversity panel of 869 biomass sorghum (Sorghum bicolor (L.) Moench) lines, which had been genotyped with 100,435 SNP markers, for plant height (PH) with biweekly measurements from 30 to 120 days after planting (DAP) and for end-of-season dry biomass yield (DBY) in four environments. We evaluated five genomic prediction models: Bayesian network (BN), Pleiotropic Bayesian network (PBN), Dynamic Bayesian network (DBN), multi-trait GBLUP (MTr-GBLUP), and multi-time GBLUP (MTi-GBLUP) models. In fivefold cross-validation, prediction accuracies ranged from 0.46 (PBN) to 0.49 (MTr-GBLUP) for DBY and from 0.47 (DBN, DAP120) to 0.75 (MTi-GBLUP, DAP60) for PH. Forward-chaining cross-validation further improved prediction accuracies of the DBN, MTi-GBLUP and MTr-GBLUP models for PH (training slice: 30-45 DAP) by 36.4-52.4% relative to the BN and PBN models. Coincidence indices (target: biomass, secondary: PH) and a coincidence index based on lines (PH time series) showed that the ranking of lines by PH changed minimally after 45 DAP. These results suggest a two-level indirect selection method for PH at harvest (first-level target trait) and DBY (second-level target trait) could be conducted earlier in the season based on ranking of lines by PH at 45 DAP (secondary trait). With the advance of high-throughput phenotyping technologies, our proposed two-level indirect selection framework could be valuable for enhancing genetic gain per unit of time when selecting on developmental traits.

Deleterious Mutation Burden and Its Association with Complex Traits in Sorghum (Sorghum bicolor).

Sorghum (Sorghum bicolor L.) is a major food cereal for millions of people worldwide. The sorghum genome, like other species, accumulates deleterious mutations, likely impacting its fitness. The lack of recombination, drift, and the coupling with favorable loci impede the removal of deleterious mutations from the genome by selection. To study how deleterious variants impact phenotypes, we identified putative deleterious mutations among ∼5.5 M segregating variants of 229 diverse biomass sorghum lines. We provide the whole-genome estimate of the deleterious burden in sorghum, showing that ∼33% of nonsynonymous substitutions are putatively deleterious. The pattern of mutation burden varies appreciably among racial groups. Across racial groups, the mutation burden correlated negatively with biomass, plant height, specific leaf area (SLA), and tissue starch content (TSC), suggesting that deleterious burden decreases trait fitness. Putatively deleterious variants explain roughly one-half of the genetic variance. However, there is only moderate improvement in total heritable variance explained for biomass (7.6%) and plant height (average of 3.1% across all stages). There is no advantage in total heritable variance for SLA and TSC. The contribution of putatively deleterious variants to phenotypic diversity therefore appears to be dependent on the genetic architecture of traits. Overall, these results suggest that incorporating putatively deleterious variants into genomic models slightly improves prediction accuracy because of extensive linkage. Knowledge of deleterious variants could be leveraged for sorghum breeding through either genome editing and/or conventional breeding that focuses on the selection of progeny with fewer deleterious alleles.

Efficiency of multi-trait, indirect, and trait-assisted genomic selection for improvement of biomass sorghum.

KEY MESSAGE: We compare genomic selection methods that use correlated traits to help predict biomass yield in sorghum, and find that trait-assisted genomic selection performs best. Genomic selection (GS) is usually performed on a single trait, but correlated traits can also help predict a focal trait through indirect or multi-trait GS. In this study, we use a pre-breeding population of biomass sorghum to compare strategies that use correlated traits to improve prediction of biomass yield, the focal trait. Correlated traits include moisture, plant height measured at monthly intervals between planting and harvest, and the area under the growth progress curve. In addition to single- and multi-trait direct and indirect GS, we test a new strategy called trait-assisted GS, in which correlated traits are used along with marker data in the validation population to predict a focal trait. Single-trait GS for biomass yield had a prediction accuracy of 0.40. Indirect GS performed best using area under the growth progress curve to predict biomass yield, with a prediction accuracy of 0.37, and did not differ from indirect multi-trait GS that also used moisture information. Multi-trait GS and single-trait GS yielded similar results, indicating that correlated traits did not improve prediction of biomass yield in a standard GS scenario. However, trait-assisted GS increased prediction accuracy by up to [Formula: see text] when using plant height in both the training and validation populations to help predict yield in the validation population. Coincidence between selected genotypes in phenotypic and genomic selection was also highest in trait-assisted GS. Overall, these results suggest that trait-assisted GS can be an efficient strategy when correlated traits are obtained earlier or more inexpensively than a focal trait.