Ridge regression and deep learning models for genome-wide selection of complex traits in New Mexican Chile peppers

BACKGROUND

Genomewide prediction estimates the genomic breeding values of selection candidates which can be utilized for population improvement and cultivar development. Ridge regression and deep learning-based selection models were implemented for yield and agronomic traits of 204 chile pepper genotypes evaluated in multi-environment trials in New Mexico, USA.

RESULTS

Accuracy of prediction differed across different models under ten-fold cross-validations, where high prediction accuracy was observed for highly heritable traits such as plant height and plant width. No model was superior across traits using 14,922 SNP markers for genomewide selection. Bayesian ridge regression had the highest average accuracy for first pod date (0.77) and total yield per plant (0.33). Multilayer perceptron (MLP) was the most superior for flowering time (0.76) and plant height (0.73), whereas the genomic BLUP model had the highest accuracy for plant width (0.62). Using a subset of 7,690 SNP loci resulting from grouping markers based on linkage disequilibrium coefficients resulted in improved accuracy for first pod date, ten pod weight, and total yield per plant, even under a relatively small training population size for MLP and random forest models. Genomic and ridge regression BLUP models were sufficient for optimal prediction accuracies for small training population size. Combining phenotypic selection and genomewide selection resulted in improved selection response for yield-related traits, indicating that integrated approaches can result in improved gains achieved through selection.

CONCLUSIONS

Accuracy values for ridge regression and deep learning prediction models demonstrate the potential of implementing genomewide selection for genetic improvement in chile pepper breeding programs. Ultimately, a large training data is relevant for improved genomic selection accuracy for the deep learning models.

Rating System for Phytophthora Root Rot Influences Quantitative Trait Locus Detection and Reveals Incomplete Dominance and Duplicative Recessive Epistatic Gene Action in Capsicum annuum

Phytophthora capsici is one of the most devastating pathogens facing pepper (Capsicum annuum) producers worldwide. Numerous factors, such as the race of the pathogen, the growing environment, and the source of resistance, have resulted in an overall lack of widely applicable molecular markers associated with resistance. Our objective was to determine the effect of the rating system on quantitative trait locus (QTL) detection and understand inheritance patterns of host resistance that can influence selection and molecular marker accuracy. We evaluated an F2:11 recombinant inbred line population screened against the highly virulent strain (Pc134) and scored using two widely used methods, developed by Bosland and Lindsey and by Black. The rating system developed by Bosland and Lindsey resulted in slightly higher logarithm of odds for the QTL on chromosome 5, and we detected a QTL on chromosome 12 uniquely using this rating system. A QTL on chromosome 10 was detected using both rating systems, but Black resulted in considerably higher logarithm of odds for this QTL compared with the Bosland and Lindsey system. Molecular markers developed were nominally better at accurately predicting the phenotype than previously published molecular markers but did not completely explain resistance in our validation populations. The inheritance pattern of resistance in one of our F2 populations did not significantly deviate from a 7:9 segregation ratio, indicating duplicative recessive epistasis. However, these results could be confounded by the presence of incomplete gene action, which was found through the improved selection accuracy when the phenotypes of heterozygous individuals were grouped with those with susceptible alleles.

Phytophthora capsici, 100 Years Later: Research Mile Markers from 1922 to 2022

In 1922, Phytophthora capsici was described by Leon Hatching Leonian as a new pathogen infecting pepper (Capsicum annuum), with disease symptoms of root rot, stem and fruit blight, seed rot, and plant wilting and death. Extensive research has been conducted on P. capsici over the last 100 years. This review succinctly describes the salient mile markers of research on P. capsici with current perspectives on the pathogen's distribution, economic importance, epidemiology, genetics and genomics, fungicide resistance, host susceptibility, pathogenicity mechanisms, and management.

Chile Pepper (Capsicum) Breeding and Improvement in the "Multi-Omics" Era

Chile pepper (Capsicum spp.) is a major culinary, medicinal, and economic crop in most areas of the world. For more than hundreds of years, chile peppers have "defined" the state of New Mexico, USA. The official state question, "Red or Green?" refers to the preference for either red or the green stage of chile pepper, respectively, reflects the value of these important commodities. The presence of major diseases, low yields, decreased acreages, and costs associated with manual labor limit production in all growing regions of the world. The New Mexico State University (NMSU) Chile Pepper Breeding Program continues to serve as a key player in the development of improved chile pepper varieties for growers and in discoveries that assist plant breeders worldwide. Among the traits of interest for genetic improvement include yield, disease resistance, flavor, and mechanical harvestability. While progress has been made, the use of conventional breeding approaches has yet to fully address producer and consumer demand for these traits in available cultivars. Recent developments in "multi-omics," that is, the simultaneous application of multiple omics approaches to study biological systems, have allowed the genetic dissection of important phenotypes. Given the current needs and production constraints, and the availability of multi-omics tools, it would be relevant to examine the application of these approaches in chile pepper breeding and improvement. In this review, we summarize the major developments in chile pepper breeding and present novel tools that can be implemented to facilitate genetic improvement. In the future, chile pepper improvement is anticipated to be more data and multi-omics driven as more advanced genetics, breeding, and phenotyping tools are developed.

Classification and Regression Models for Genomic Selection of Skewed Phenotypes: A Case for Disease Resistance in Winter Wheat (Triticum aestivum L.)

Most genomic prediction models are linear regression models that assume continuous and normally distributed phenotypes, but responses to diseases such as stripe rust (caused by Puccinia striiformis f. sp. tritici) are commonly recorded in ordinal scales and percentages. Disease severity (SEV) and infection type (IT) data in germplasm screening nurseries generally do not follow these assumptions. On this regard, researchers may ignore the lack of normality, transform the phenotypes, use generalized linear models, or use supervised learning algorithms and classification models with no restriction on the distribution of response variables, which are less sensitive when modeling ordinal scores. The goal of this research was to compare classification and regression genomic selection models for skewed phenotypes using stripe rust SEV and IT in winter wheat. We extensively compared both regression and classification prediction models using two training populations composed of breeding lines phenotyped in 4 years (2016–2018 and 2020) and a diversity panel phenotyped in 4 years (2013–2016). The prediction models used 19,861 genotyping-by-sequencing single-nucleotide polymorphism markers. Overall, square root transformed phenotypes using ridge regression best linear unbiased prediction and support vector machine regression models displayed the highest combination of accuracy and relative efficiency across the regression and classification models. Furthermore, a classification system based on support vector machine and ordinal Bayesian models with a 2-Class scale for SEV reached the highest class accuracy of 0.99. This study showed that breeders can use linear and non-parametric regression models within their own breeding lines over combined years to accurately predict skewed phenotypes.

Genomic regions and candidate genes linked with Phytophthora capsici root rot resistance in chile pepper (Capsicum annuum L.)

BACKGROUND

Phytophthora root rot, caused by Phytophthora capsici, is a major disease affecting Capsicum production worldwide. A recombinant inbred line (RIL) population derived from the hybridization between 'Criollo de Morellos-334' (CM-334), a resistant landrace from Mexico, and 'Early Jalapeno', a susceptible cultivar was genotyped using genotyping-by-sequencing (GBS)-derived single nucleotide polymorphism (SNP) markers. A GBS-SNP based genetic linkage map for the RIL population was constructed. Quantitative trait loci (QTL) mapping dissected the genetic architecture of P. capsici resistance and candidate genes linked to resistance for this important disease were identified.

RESULTS

Development of a genetic linkage map using 1,973 GBS-derived polymorphic SNP markers identified 12 linkage groups corresponding to the 12 chromosomes of chile pepper, with a total length of 1,277.7 cM and a marker density of 1.5 SNP/cM. The maximum gaps between consecutive SNP markers ranged between 1.9 (LG7) and 13.5 cM (LG5). Collinearity between genetic and physical positions of markers reached a maximum of 0.92 for LG8. QTL mapping identified genomic regions associated with P. capsici resistance in chromosomes P5, P8, and P9 that explained between 19.7 and 30.4% of phenotypic variation for resistance. Additive interactions between QTL in chromosomes P5 and P8 were observed. The role of chromosome P5 as major genomic region containing P. capsici resistance QTL was established. Through candidate gene analysis, biological functions associated with response to pathogen infections, regulation of cyclin-dependent protein serine/threonine kinase activity, and epigenetic mechanisms such as DNA methylation were identified.

CONCLUSIONS

Results support the genetic complexity of the P. capsici-Capsicum pathosystem and the possible role of epigenetics in conferring resistance to Phytophthora root rot. Significant genomic regions and candidate genes associated with disease response and gene regulatory activity were identified which allows for a deeper understanding of the genomic landscape of Phytophthora root rot resistance in chile pepper.

Gains through selection for grain yield in a winter wheat breeding program

Increased genetic gain for complex traits in plant breeding programs can be achieved through different selection strategies. The objective of this study was to compare potential gains for grain yield in a winter wheat breeding program through estimating response to selection R values across several selection approaches including phenotypic (PS), marker-based (MS), genomic (GS), and a combination of PS and GS (PS+GS). Ten populations of Washington State University (WSU) winter wheat breeding lines including a diversity panel and F5 and double haploid lines evaluated from 2015 to 2019 growing seasons for grain yield in Lind and Pullman, WA, USA were used in the study. Selection was conducted by selecting the top 20% of lines based on observed yield (PS strategy), genomic estimated breeding values (GS), presence of yield "enhancing" alleles of the most significant single nucleotide polymorphism (SNP) markers identified from genome-wide association mapping (MS), and high observed yield and estimated breeding values (PS+GS). Overall, PS compared to other individual selection strategies (MS and GS) showed the highest mean response (R = 0.61) within the same environment. When combined with GS, a 23% improvement in R for yield was observed, indicating that gains could be improved by complementing traditional PS with GS within the same environment. Validating selection strategies in different environments resulted in low to negative R values indicating the effects of genotype-by-environment interactions for grain yield. MS was not successful in terms of R relative to the other selection approaches; using this strategy resulted in a significant (P < 0.05) decrease in response to selection compared with the other approaches. An integrated PS+GS approach could result in optimal genetic gain within the same environment, whereas a PS strategy might be a viable option for grain yield validated in different environments. Altogether, we demonstrated that gains through increased response to selection for yield could be achieved in the WSU winter wheat breeding program by implementing different selection strategies either exclusively or in combination.

Deep Learning for Predicting Complex Traits in Spring Wheat Breeding Program

Genomic selection (GS) is transforming the field of plant breeding and implementing models that improve prediction accuracy for complex traits is needed. Analytical methods for complex datasets traditionally used in other disciplines represent an opportunity for improving prediction accuracy in GS. Deep learning (DL) is a branch of machine learning (ML) which focuses on densely connected networks using artificial neural networks for training the models. The objective of this research was to evaluate the potential of DL models in the Washington State University spring wheat breeding program. We compared the performance of two DL algorithms, namely multilayer perceptron (MLP) and convolutional neural network (CNN), with ridge regression best linear unbiased predictor (rrBLUP), a commonly used GS model. The dataset consisted of 650 recombinant inbred lines (RILs) from a spring wheat nested association mapping (NAM) population planted from 2014-2016 growing seasons. We predicted five different quantitative traits with varying genetic architecture using cross-validations (CVs), independent validations, and different sets of SNP markers. Hyperparameters were optimized for DL models by lowering the root mean square in the training set, avoiding model overfitting using dropout and regularization. DL models gave 0 to 5% higher prediction accuracy than rrBLUP model under both cross and independent validations for all five traits used in this study. Furthermore, MLP produces 5% higher prediction accuracy than CNN for grain yield and grain protein content. Altogether, DL approaches obtained better prediction accuracy for each trait, and should be incorporated into a plant breeder's toolkit for use in large scale breeding programs.

Genomic Prediction and Indirect Selection for Grain Yield in US Pacific Northwest Winter Wheat Using Spectral Reflectance Indices from High-Throughput Phenotyping

Secondary traits from high-throughput phenotyping could be used to select for complex target traits to accelerate plant breeding and increase genetic gains. This study aimed to evaluate the potential of using spectral reflectance indices (SRI) for indirect selection of winter-wheat lines with high yield potential and to assess the effects of including secondary traits on the prediction accuracy for yield. A total of five SRIs were measured in a diversity panel, and F5 and doubled haploid wheat breeding populations planted between 2015 and 2018 in Lind and Pullman, WA. The winter-wheat panels were genotyped with 11,089 genotyping-by-sequencing derived markers. Spectral traits showed moderate to high phenotypic and genetic correlations, indicating their potential for indirect selection of lines with high yield potential. Inclusion of correlated spectral traits in genomic prediction models resulted in significant (p < 0.001) improvement in prediction accuracy for yield. Relatedness between training and test populations and heritability were among the principal factors affecting accuracy. Our results demonstrate the potential of using spectral indices as proxy measurements for selecting lines with increased yield potential and for improving prediction accuracy to increase genetic gains for complex traits in US Pacific Northwest winter wheat.

Accuracy of genomic selection for grain yield and agronomic traits in soft red winter wheat

BACKGROUND

Genomic selection has the potential to increase genetic gains by using molecular markers as predictors of breeding values of individuals. This study evaluated the accuracy of predictions for grain yield, heading date, plant height, and yield components in soft red winter wheat under different prediction scenarios. Response to selection for grain yield was also compared across different selection strategies- phenotypic, marker-based, genomic, combination of phenotypic and genomic, and random selections.

RESULTS

Genomic selection was implemented through a ridge regression best linear unbiased prediction model in two scenarios- cross-validations and independent predictions. Accuracy for cross-validations was assessed using a diverse panel under different marker number, training population size, relatedness between training and validation populations, and inclusion of fixed effect in the model. The population in the first scenario was then trained and used to predict grain yield of biparental populations for independent validations. Using subsets of significant markers from association mapping increased accuracy by 64-70% for grain yield but resulted in lower accuracy for traits with high heritability such as plant height. Increasing size of training population resulted in an increase in accuracy, with maximum values reached when ~ 60% of the lines were used as a training panel. Predictions using related subpopulations also resulted in higher accuracies. Inclusion of major growth habit genes as fixed effect in the model caused increase in grain yield accuracy under a cross-validation procedure. Independent predictions resulted in accuracy ranging between - 0.14 and 0.43, dependent on the grouping of site-year data for the training and validation populations. Genomic selection was "superior" to marker-based selection in terms of response to selection for yield. Supplementing phenotypic with genomic selection resulted in approximately 10% gain in response compared to using phenotypic selection alone.

CONCLUSIONS

Our results showed the effects of different factors on accuracy for yield and agronomic traits. Among the factors studied, training population size and relatedness between training and validation population had the greatest impact on accuracy. Ultimately, combining phenotypic with genomic selection would be relevant for accelerating genetic gains for yield in winter wheat.