Genomic selection: genome-wide prediction in plant improvement

Machine learning algorithms translate big data into predictive breeding accuracy

Statistical machine learning (ML) extracts patterns from extensive genomic, phenotypic, and environmental data. ML algorithms automatically identify relevant features and use cross-validation to ensure robust models and improve prediction reliability in new lines. Furthermore, ML analyses of genotype-by-environment (G×E) interactions can offer insights into the genetic factors that affect performance in specific environments. By leveraging historical breeding data, ML streamlines strategies and automates analyses to reveal genomic patterns. In this review we examine the transformative impact of big data, including multi-trait genomics, phenomics, and environmental covariables, on genomic-enabled prediction in plant breeding. We discuss how big data and ML are revolutionizing the field by enhancing prediction accuracy, deepening our understanding of G×E interactions, and optimizing breeding strategies through the analysis of extensive and diverse datasets.

Regulatory balance between ear rot resistance and grain yield and their breeding applications in maize and other crops

BACKGROUND

Fungi are prevalent pathogens that cause substantial yield losses of major crops. Ear rot (ER), which is primarily induced by Fusarium or Aspergillus species, poses a significant challenge to maize production worldwide. ER resistance is regulated by several small effect quantitative trait loci (QTLs). To date, only a few ER-related genes have been identified that impede molecular breeding efforts to breed ER-resistant maize varieties.

AIM OF REVIEW

Our aim here is to explore the research progress and mine genic resources related to ER resistance, and to propose a regulatory model elucidating the ER-resistant mechanism in maize as well as a trade-off model illustrating how crops balance fungal resistance and grain yield. Key Scientific Concepts of Review: This review presents a comprehensive bibliometric analysis of the research history and current trends in the genetic and molecular regulation underlying ER resistance in maize. Moreover, we analyzed and discovered the genic resources by identifying 162 environmentally stable loci (ESLs) from various independent forward genetics studies as well as 1391 conservatively differentially expressed genes (DEGs) that respond to Fusarium or Aspergillus infection through multi-omics data analysis. Additionally, this review discusses the syntenies found among maize ER, wheat Fusariumhead blight (FHB), and rice Bakanaedisease (RBD) resistance-related loci, along with the significant overlap between fungal resistance loci and reported yield-related loci, thus providing valuable insights into the regulatory mechanisms underlying the trade-offs between yield and defense in crops.

Incremental Inverse Design of Desired Soybean Phenotypes

We present an application of computational inverse design, which reverses the conventional trial-and-error forward design paradigm, optimizes biological phenotype by directly modifying genotype. The limitations of inverse design in genotype-to-bulk phenotype (G-BP) mapping can be addressed via an established design paradigm: "design, build, test, learn" (DBTL), where computational inverse design automates both the design and learn phases. In any context, inverse design is limited by the fundamental "one-to-many" nature of the inverse function. G-BP inverse design is further limited by the number of single nucleotide polymorphisms that can be made to a member of the population while maintaining feasibility of genotype creation and biological viability. Considering these limitations, we propose a design paradigm based on incremental optimization of phenotype through a combined computational and experimental approach. We intend this work to be a foundational synthesis of well-known techniques applied to the context of genotype-to-bulk phenotype inverse design, which has not yet been performed in the literature. The design pipeline can optimize phenotype by either directly proposing genotypic changes, or simply by suggesting parents to be used for selective breeding. The soybean nested association matrix data set is used to present an in silico case study of the design pipeline by performing optimization that maximizes protein content while constraining other phenotypes. A random forest (RF) is used to model the genotype-to-phenotype relationship, and a genetic algorithm is used to query the RF until a feasible genotype with desired phenotype is discovered. After 20 in silico DBTL cycles, a final population of individuals with a mean protein content of 36.13%, an increase of three standard deviations above the original mean is suggested.

Integrating targeted genetic markers to genotyping-by-sequencing for an ultimate genotyping tool

New selection methods, using trait-specific markers (marker-assisted selection (MAS)) and/or genome-wide markers (genomic selection (GS)), are becoming increasingly widespread in breeding programs. This new era requires innovative and cost-efficient solutions for genotyping. Reduction in sequencing cost has enhanced the use of high-throughput low-cost genotyping methods such as genotyping-by-sequencing (GBS) for genome-wide single-nucleotide polymorphism (SNP) profiling in large breeding populations. However, the major weakness of GBS methodologies is their inability to genotype targeted markers. Conversely, targeted methods, such as amplicon sequencing (AmpSeq), often face cost constraints, hindering genome-wide genotyping across a large cohort. Although GBS and AmpSeq data can be generated from the same sample, an efficient method to achieve this is lacking. In this study, we present the Genome-wide & Targeted Amplicon (GTA) genotyping platform, an innovative way to integrate multiplex targeted amplicons into the GBS library preparation to provide an all-in-one cost-effective genotyping solution to breeders and research communities. Custom primers were designed to target 23 and 36 high-value markers associated with key agronomical traits in soybean and barley, respectively. The resulting multiplex amplicons were compatible with the GBS library preparation enabling both GBS and targeted genotyping data to be produced efficiently and cost-effectively. To facilitate data analysis, we have introduced Fast-GBS.v3, a user-friendly bioinformatic pipeline that generates comprehensive outputs from data obtained following sequencing of GTA libraries. This high-throughput low-cost approach will greatly facilitate the application of DNA markers as it provides required markers for both MAS and GS in a single assay.

Epistasis and pleiotropy-induced variation for plant breeding

Epistasis refers to nonallelic interaction between genes that cause bias in estimates of genetic parameters for a phenotype with interactions of two or more genes affecting the same trait. Partitioning of epistatic effects allows true estimation of the genetic parameters affecting phenotypes. Multigenic variation plays a central role in the evolution of complex characteristics, among which pleiotropy, where a single gene affects several phenotypic characters, has a large influence. While pleiotropic interactions provide functional specificity, they increase the challenge of gene discovery and functional analysis. Overcoming pleiotropy-based phenotypic trade-offs offers potential for assisting breeding for complex traits. Modelling higher order nonallelic epistatic interaction, pleiotropy and non-pleiotropy-induced variation, and genotype × environment interaction in genomic selection may provide new paths to increase the productivity and stress tolerance for next generation of crop cultivars. Advances in statistical models, software and algorithm developments, and genomic research have facilitated dissecting the nature and extent of pleiotropy and epistasis. We overview emerging approaches to exploit positive (and avoid negative) epistatic and pleiotropic interactions in a plant breeding context, including developing avenues of artificial intelligence, novel exploitation of large-scale genomics and phenomics data, and involvement of genes with minor effects to analyse epistatic interactions and pleiotropic quantitative trait loci, including missing heritability.

Global Genotype by Environment Prediction Competition Reveals That Diverse Modeling Strategies Can Deliver Satisfactory Maize Yield Estimates

Predicting phenotypes from a combination of genetic and environmental factors is a grand challenge of modern biology. Slight improvements in this area have the potential to save lives, improve food and fuel security, permit better care of the planet, and create other positive outcomes. In 2022 and 2023 the first open-to-the-public Genomes to Fields (G2F) initiative Genotype by Environment (GxE) prediction competition was held using a large dataset including genomic variation, phenotype and weather measurements and field management notes, gathered by the project over nine years. The competition attracted registrants from around the world with representation from academic, government, industry, and non-profit institutions as well as unaffiliated. These participants came from diverse disciplines include plant science, animal science, breeding, statistics, computational biology and others. Some participants had no formal genetics or plant-related training, and some were just beginning their graduate education. The teams applied varied methods and strategies, providing a wealth of modeling knowledge based on a common dataset. The winner's strategy involved two models combining machine learning and traditional breeding tools: one model emphasized environment using features extracted by Random Forest, Ridge Regression and Least-squares, and one focused on genetics. Other high-performing teams' methods included quantitative genetics, classical machine learning/deep learning, mechanistic models, and model ensembles. The dataset factors used, such as genetics; weather; and management data, were also diverse, demonstrating that no single model or strategy is far superior to all others within the context of this competition.

PNNGS, a multi-convolutional parallel neural network for genomic selection

Genomic selection (GS) can accomplish breeding faster than phenotypic selection. Improving prediction accuracy is the key to promoting GS. To improve the GS prediction accuracy and stability, we introduce parallel convolution to deep learning for GS and call it a parallel neural network for genomic selection (PNNGS). In PNNGS, information passes through convolutions of different kernel sizes in parallel. The convolutions in each branch are connected with residuals. Four different Lp loss functions train PNNGS. Through experiments, the optimal number of parallel paths for rice, sunflower, wheat, and maize is found to be 4, 6, 4, and 3, respectively. Phenotype prediction is performed on 24 cases through ridge-regression best linear unbiased prediction (RRBLUP), random forests (RF), support vector regression (SVR), deep neural network genomic prediction (DNNGP), and PNNGS. Serial DNNGP and parallel PNNGS outperform the other three algorithms. On average, PNNGS prediction accuracy is 0.031 larger than DNNGP prediction accuracy, indicating that parallelism can improve the GS model. Plants are divided into clusters through principal component analysis (PCA) and K-means clustering algorithms. The sample sizes of different clusters vary greatly, indicating that this is unbalanced data. Through stratified sampling, the prediction stability and accuracy of PNNGS are improved. When the training samples are reduced in small clusters, the prediction accuracy of PNNGS decreases significantly. Increasing the sample size of small clusters is critical to improving the prediction accuracy of GS.

Genomic selection for crop improvement in fruits and vegetables: a systematic scoping review

To ensure the nutritional needs of an expanding global population, it is crucial to optimize the growing capabilities and breeding values of fruit and vegetable crops. While genomic selection, initially implemented in animal breeding, holds tremendous potential, its utilization in fruit and vegetable crops remains underexplored. In this systematic review, we reviewed 63 articles covering genomic selection and its applications across 25 different types of fruit and vegetable crops over the last decade. The traits examined were directly related to the edible parts of the crops and carried significant economic importance. Comparative analysis with WHO/FAO data identified potential economic drivers underlying the study focus of some crops and highlighted crops with potential for further genomic selection research and application. Factors affecting genomic selection accuracy in fruit and vegetable studies are discussed and suggestions made to assist in their implementation into plant breeding schemes. Genetic gain in fruits and vegetables can be improved by utilizing genomic selection to improve selection intensity, accuracy, and integration of genetic variation. However, the reduction of breeding cycle times may not be beneficial in crops with shorter life cycles such as leafy greens as compared to fruit trees. There is an urgent need to integrate genomic selection methods into ongoing breeding programs and assess the actual genomic estimated breeding values of progeny resulting from these breeding programs against the prediction models.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11032-024-01497-2.

Genomic prediction and genome-wide association study using combined genotypic data from different genotyping systems: application to apple fruit quality traits

With advances in next-generation sequencing technologies, various marker genotyping systems have been developed for genomics-based approaches such as genomic selection (GS) and genome-wide association study (GWAS). As new genotyping platforms are developed, data from different genotyping platforms must be combined. However, the potential use of combined data for GS and GWAS has not yet been clarified. In this study, the accuracy of genomic prediction (GP) and the detection power of GWAS increased for most fruit quality traits of apples when using combined data from different genotyping systems, Illumina Infinium single-nucleotide polymorphism array and genotyping by random amplicon sequencing-direct (GRAS-Di) systems. In addition, the GP model, which considered the inbreeding effect, further improved the accuracy of the seven fruit traits. Runs of homozygosity (ROH) islands overlapped with the significantly associated regions detected by the GWAS for several fruit traits. Breeders may have exploited these regions to select promising apples by breeders, increasing homozygosity. These results suggest that combining genotypic data from different genotyping platforms benefits the GS and GWAS of fruit quality traits in apples. Information on inbreeding could be beneficial for improving the accuracy of GS for fruit traits of apples; however, further analysis is required to elucidate the relationship between the fruit traits and inbreeding depression (e.g. decreased vigor).

The potential of genome editing to create novel alleles of resistance genes in rice

Rice, a staple food for a significant portion of the global population, faces persistent threats from various pathogens and pests, necessitating the development of resilient crop varieties. Deployment of resistance genes in rice is the best practice to manage diseases and reduce environmental damage by reducing the application of agro-chemicals. Genome editing technologies, such as CRISPR-Cas, have revolutionized the field of molecular biology, offering precise and efficient tools for targeted modifications within the rice genome. This study delves into the application of these tools to engineer novel alleles of resistance genes in rice, aiming to enhance the plant's innate ability to combat evolving threats. By harnessing the power of genome editing, researchers can introduce tailored genetic modifications that bolster the plant's defense mechanisms without compromising its essential characteristics. In this study, we synthesize recent advancements in genome editing methodologies applicable to rice and discuss the ethical considerations and regulatory frameworks surrounding the creation of genetically modified crops. Additionally, it explores potential challenges and future prospects for deploying edited rice varieties in agricultural landscapes. In summary, this study highlights the promise of genome editing in reshaping the genetic landscape of rice to confront emerging challenges, contributing to global food security and sustainable agriculture practices.

Tools and Techniques to Accelerate Crop Breeding

As climate changes and a growing global population continue to escalate the need for greater production capabilities of food crops, technological advances in agricultural and crop research will remain a necessity. While great advances in crop improvement over the past century have contributed to massive increases in yield, classic breeding schemes lack the rate of genetic gain needed to meet future demands. In the past decade, new breeding techniques and tools have been developed to aid in crop improvement. One such advancement is the use of speed breeding. Speed breeding is known as the application of methods that significantly reduce the time between crop generations, thereby streamlining breeding and research efforts. These rapid-generation advancement tactics help to accelerate the pace of crop improvement efforts to sustain food security and meet the food, feed, and fiber demands of the world's growing population. Speed breeding may be achieved through a variety of techniques, including environmental optimization, genomic selection, CRISPR-Cas9 technology, and epigenomic tools. This review aims to discuss these prominent advances in crop breeding technologies and techniques that have the potential to greatly improve plant breeders' ability to rapidly produce vital cultivars.

Low-density SNP markers with high prediction accuracy of genomic selection for bacterial wilt resistance in tomato

Bacterial wilt (BW) is a soil-borne disease that leads to severe damage in tomato. Host resistance against BW is considered polygenic and effective in controlling this destructive disease. In this study, genomic selection (GS), which is a promising breeding strategy to improve quantitative traits, was investigated for BW resistance. Two tomato collections, TGC1 (n = 162) and TGC2 (n = 191), were used as training populations. Disease severity was assessed using three seedling assays in each population, and the best linear unbiased prediction (BLUP) values were obtained. The 31,142 SNP data were generated using the 51K Axiom array™ in the training populations. With these data, six GS models were trained to predict genomic estimated breeding values (GEBVs) in three populations (TGC1, TGC2, and combined). The parametric models Bayesian LASSO and RR-BLUP resulted in higher levels of prediction accuracy compared with all the non-parametric models (RKHS, SVM, and random forest) in two training populations. To identify low-density markers, two subsets of 1,557 SNPs were filtered based on marker effects (Bayesian LASSO) and variable importance values (random forest) in the combined population. An additional subset was generated using 1,357 SNPs from a genome-wide association study. These subsets showed prediction accuracies of 0.699 to 0.756 in Bayesian LASSO and 0.670 to 0.682 in random forest, which were higher relative to the 31,142 SNPs (0.625 and 0.614). Moreover, high prediction accuracies (0.743 and 0.702) were found with a common set of 135 SNPs derived from the three subsets. The resulting low-density SNPs will be useful to develop a cost-effective GS strategy for BW resistance in tomato breeding programs.

Feature engineering of environmental covariates improves plant genomic-enabled prediction

Introduction

Because Genomic selection (GS) is a predictive methodology, it needs to guarantee high-prediction accuracies for practical implementations. However, since many factors affect the prediction performance of this methodology, its practical implementation still needs to be improved in many breeding programs. For this reason, many strategies have been explored to improve the prediction performance of this methodology.

Methods

When environmental covariates are incorporated as inputs in the genomic prediction models, this information only sometimes helps increase prediction performance. For this reason, this investigation explores the use of feature engineering on the environmental covariates to enhance the prediction performance of genomic prediction models.

Results and discussion

We found that across data sets, feature engineering helps reduce prediction error regarding only the inclusion of the environmental covariates without feature engineering by 761.625% across predictors. These results are very promising regarding the potential of feature engineering to enhance prediction accuracy. However, since a significant gain in prediction accuracy was observed in only some data sets, further research is required to guarantee a robust feature engineering strategy to incorporate the environmental covariates.

Review of applications of artificial intelligence (AI) methods in crop research

Sophisticated and modern crop improvement techniques can bridge the gap for feeding the ever-increasing population. Artificial intelligence (AI) refers to the simulation of human intelligence in machines, which refers to the application of computational algorithms, machine learning (ML) and deep learning (DL) techniques. This is aimed to generalise patterns and relationships from historical data, employing various mathematical optimisation techniques thus making prediction models for facilitating selection of superior genotypes. These techniques are less resource intensive and can solve the problem based on the analysis of large-scale phenotypic datasets. ML for genomic selection (GS) uses high-throughput genotyping technologies to gather genetic information on a large number of markers across the genome. The prediction of GS models is based on the mathematical relation between genotypic and phenotypic data from the training population. ML techniques have emerged as powerful tools for genome editing through analysing large-scale genomic data and facilitating the development of accurate prediction models. Precise phenotyping is a prerequisite to advance crop breeding for solving agricultural production-related issues. ML algorithms can solve this problem through generating predictive models, based on the analysis of large-scale phenotypic datasets. DL models also have the potential reliability of precise phenotyping. This review provides a comprehensive overview on various ML and DL models, their applications, potential to enhance the efficiency, specificity and safety towards advanced crop improvement protocols such as genomic selection, genome editing, along with phenotypic prediction to promote accelerated breeding.

Genomic selection in plant breeding: Key factors shaping two decades of progress

Genomic selection, the application of genomic prediction (GP) models to select candidate individuals, has significantly advanced in the past two decades, effectively accelerating genetic gains in plant breeding. This article provides a holistic overview of key factors that have influenced GP in plant breeding during this period. We delved into the pivotal roles of training population size and genetic diversity, and their relationship with the breeding population, in determining GP accuracy. Special emphasis was placed on optimizing training population size. We explored its benefits and the associated diminishing returns beyond an optimum size. This was done while considering the balance between resource allocation and maximizing prediction accuracy through current optimization algorithms. The density and distribution of single-nucleotide polymorphisms, level of linkage disequilibrium, genetic complexity, trait heritability, statistical machine-learning methods, and non-additive effects are the other vital factors. Using wheat, maize, and potato as examples, we summarize the effect of these factors on the accuracy of GP for various traits. The search for high accuracy in GP-theoretically reaching one when using the Pearson's correlation as a metric-is an active research area as yet far from optimal for various traits. We hypothesize that with ultra-high sizes of genotypic and phenotypic datasets, effective training population optimization methods and support from other omics approaches (transcriptomics, metabolomics and proteomics) coupled with deep-learning algorithms could overcome the boundaries of current limitations to achieve the highest possible prediction accuracy, making genomic selection an effective tool in plant breeding.

Comparison of ddRADseq and EUChip60K SNP genotyping systems for population genetics and genomic selection in Eucalyptus dunnii (Maiden)

Eucalyptus dunnii is one of the most important Eucalyptus species for short-fiber pulp production in regions where other species of the genus are affected by poor soil and climatic conditions. In this context, E. dunnii holds promise as a resource to address and adapt to the challenges of climate change. Despite its rapid growth and favorable wood properties for solid wood products, the advancement of its improvement remains in its early stages. In this work, we evaluated the performance of two single nucleotide polymorphism, (SNP), genotyping methods for population genetics analysis and Genomic Selection in E. dunnii. Double digest restriction-site associated DNA sequencing (ddRADseq) was compared with the EUChip60K array in 308 individuals from a provenance-progeny trial. The compared SNP set included 8,011 and 19,008 informative SNPs distributed along the 11 chromosomes, respectively. Although the two datasets differed in the percentage of missing data, genome coverage, minor allele frequency and estimated genetic diversity parameters, they revealed a similar genetic structure, showing two subpopulations with little differentiation between them, and low linkage disequilibrium. GS analyses were performed for eleven traits using Genomic Best Linear Unbiased Prediction (GBLUP) and a conventional pedigree-based model (ABLUP). Regardless of the SNP dataset, the predictive ability (PA) of GBLUP was better than that of ABLUP for six traits (Cellulose content, Total and Ethanolic extractives, Total and Klason lignin content and Syringyl and Guaiacyl lignin monomer ratio). When contrasting the SNP datasets used to estimate PAs, the GBLUP-EUChip60K model gave higher and significant PA values for six traits, meanwhile, the values estimated using ddRADseq gave higher values for three other traits. The PAs correlated positively with narrow sense heritabilities, with the highest correlations shown by the ABLUP and GBLUP-EUChip60K. The two genotyping methods, ddRADseq and EUChip60K, are generally comparable for population genetics and genomic prediction, demonstrating the utility of the former when subjected to rigorous SNP filtering. The results of this study provide a basis for future whole-genome studies using ddRADseq in non-model forest species for which SNP arrays have not yet been developed.

Ensemble learning for integrative prediction of genetic values with genomic variants

BACKGROUND

Whole genome variants offer sufficient information for genetic prediction of human disease risk, and prediction of animal and plant breeding values. Many sophisticated statistical methods have been developed for enhancing the predictive ability. However, each method has its own advantages and disadvantages, so far, no one method can beat others.

RESULTS

We herein propose an Ensemble Learning method for Prediction of Genetic Values (ELPGV), which assembles predictions from several basic methods such as GBLUP, BayesA, BayesB and BayesCπ, to produce more accurate predictions. We validated ELPGV with a variety of well-known datasets and a serious of simulated datasets. All revealed that ELPGV was able to significantly enhance the predictive ability than any basic methods, for instance, the comparison p-value of ELPGV over basic methods were varied from 4.853E-118 to 9.640E-20 for WTCCC dataset.

CONCLUSIONS

ELPGV is able to integrate the merit of each method together to produce significantly higher predictive ability than any basic methods and it is simple to implement, fast to run, without using genotype data. is promising for wide application in genetic predictions.

Genome-wide association study and genomic prediction for resistance to brown planthopper in rice

The brown planthopper (BPH) is the most destructive insect pest that threatens rice production globally. Developing rice varieties incorporating BPH-resistant genes has proven to be an effective control measure against BPH. In this study, we assessed the resistance of a core collection consisting of 502 rice germplasms by evaluating resistance scores, weight gain rates and honeydew excretions. A total of 117 rice varieties (23.31%) exhibited resistance to BPH. Genome-wide association studies (GWAS) were performed on both the entire panel of 502 rice varieties and its subspecies, and 6 loci were significantly associated with resistance scores (P value < 1.0e-8). Within these loci, we identified eight candidate genes encoding receptor-like protein kinase (RLK), nucleotide-binding and leucine-rich repeat (NB-LRR), or LRR proteins. Two loci had not been detected in previous study and were entirely novel. Furthermore, we evaluated the predictive ability of genomic selection for resistance to BPH. The results revealed that the highest prediction accuracy for BPH resistance reached 0.633. As expected, the prediction accuracy increased progressively with an increasing number of SNPs, and a total of 6.7K SNPs displayed comparable accuracy to 268K SNPs. Among various statistical models tested, the random forest model exhibited superior predictive accuracy. Moreover, increasing the size of training population improved prediction accuracy; however, there was no significant difference in prediction accuracy between a training population size of 737 and 1179. Additionally, when there existed close genetic relatedness between the training and validation populations, higher prediction accuracies were observed compared to scenarios when they were genetically distant. These findings provide valuable resistance candidate genes and germplasm resources and are crucial for the application of genomic selection for breeding durable BPH-resistant rice varieties.

The effect of marker types and density on genomic prediction and GWAS of key performance traits in tetraploid potato

Genomic prediction and genome-wide association studies are becoming widely employed in potato key performance trait QTL identifications and to support potato breeding using genomic selection. Elite cultivars are tetraploid and highly heterozygous but also share many common ancestors and generation-spanning inbreeding events, resulting from the clonal propagation of potatoes through seed potatoes. Consequentially, many SNP markers are not in a 1:1 relationship with a single allele variant but shared over several alleles that might exert varying effects on a given trait. The impact of such redundant "diluted" predictors on the statistical models underpinning genome-wide association studies (GWAS) and genomic prediction has scarcely been evaluated despite the potential impact on model accuracy and performance. We evaluated the impact of marker location, marker type, and marker density on the genomic prediction and GWAS of five key performance traits in tetraploid potato (chipping quality, dry matter content, length/width ratio, senescence, and yield). A 762-offspring panel of a diallel cross of 18 elite cultivars was genotyped by sequencing, and markers were annotated according to a reference genome. Genomic prediction models (GBLUP) were trained on four marker subsets [non-synonymous (29,553 SNPs), synonymous (31,229), non-coding (32,388), and a combination], and robustness to marker reduction was investigated. Single-marker regression GWAS was performed for each trait and marker subset. The best cross-validated prediction correlation coefficients of 0.54, 0.75, 0.49, 0.35, and 0.28 were obtained for chipping quality, dry matter content, length/width ratio, senescence, and yield, respectively. The trait prediction abilities were similar across all marker types, with only non-synonymous variants improving yield predictive ability by 16%. Marker reduction response did not depend on marker type but rather on trait. Traits with high predictive abilities, e.g., dry matter content, reached a plateau using fewer markers than traits with intermediate-low correlations, such as yield. The predictions were unbiased across all traits, marker types, and all marker densities >100 SNPs. Our results suggest that using non-synonymous variants does not enhance the performance of genomic prediction of most traits. The major known QTLs were identified by GWAS and were reproducible across exonic and whole-genome variant sets for dry matter content, length/width ratio, and senescence. In contrast, minor QTL detection was marker type dependent.

Designing future peanut: the power of genomics-assisted breeding

KEY MESSAGE

Integrating GAB methods with high-throughput phenotyping, genome editing, and speed breeding hold great potential in designing future smart peanut cultivars to meet market and food supply demands. Cultivated peanut (Arachis hypogaea L.), a legume crop greatly valued for its nourishing food, cooking oil, and fodder, is extensively grown worldwide. Despite decades of classical breeding efforts, the actual on-farm yield of peanut remains below its potential productivity due to the complicated interplay of genotype, environment, and management factors, as well as their intricate interactions. Integrating modern genomics tools into crop breeding is necessary to fast-track breeding efficiency and rapid progress. When combined with speed breeding methods, this integration can substantially accelerate the breeding process, leading to faster access of improved varieties to farmers. Availability of high-quality reference genomes for wild diploid progenitors and cultivated peanuts has accelerated the process of gene/quantitative locus discovery, developing markers and genotyping assays as well as a few molecular breeding products with improved resistance and oil quality. The use of new breeding tools, e.g., genomic selection, haplotype-based breeding, speed breeding, high-throughput phenotyping, and genome editing, is probable to boost genetic gains in peanut. Moreover, renewed attention to efficient selection and exploitation of targeted genetic resources is also needed to design high-quality and high-yielding peanut cultivars with main adaptation attributes. In this context, the combination of genomics-assisted breeding (GAB), genome editing, and speed breeding hold great potential in designing future improved peanut cultivars to meet market and food supply demands.

Genomic prediction for agronomic traits in a diverse Flax (Linum usitatissimum L.) germplasm collection

Breeding programs require exhaustive phenotyping of germplasms, which is time-demanding and expensive. Genomic prediction helps breeders harness the diversity of any collection to bypass phenotyping. Here, we examined the genomic prediction's potential for seed yield and nine agronomic traits using 26,171 single nucleotide polymorphism (SNP) markers in a set of 337 flax (Linum usitatissimum L.) germplasm, phenotyped in five environments. We evaluated 14 prediction models and several factors affecting predictive ability based on cross-validation schemes. Models yielded significant variation among predictive ability values across traits for the whole marker set. The ridge regression (RR) model covering additive gene action yielded better predictive ability for most of the traits, whereas it was higher for low heritable traits by models capturing epistatic gene action. Marker subsets based on linkage disequilibrium decay distance gave significantly higher predictive abilities to the whole marker set, but for randomly selected markers, it reached a plateau above 3000 markers. Markers having significant association with traits improved predictive abilities compared to the whole marker set when marker selection was made on the whole population instead of the training set indicating a clear overfitting. The correction for population structure did not increase predictive abilities compared to the whole collection. However, stratified sampling by picking representative genotypes from each cluster improved predictive abilities. The indirect predictive ability for a trait was proportionate to its correlation with other traits. These results will help breeders to select the best models, optimum marker set, and suitable genotype set to perform an indirect selection for quantitative traits in this diverse flax germplasm collection.

Genetic gains in IRRI's rice salinity breeding and elite panel development as a future breeding resource

KEY MESSAGE

Estimating genetic gains and formulating a future salinity elite breeding panel for rice pave the way for developing better high-yielding salinity tolerant lines with enhanced genetic gains. Genetic gain is a crucial parameter to check the breeding program's success and help optimize future breeding strategies for enhanced genetic gains. To estimate the genetic gains in IRRI's salinity breeding program and identify the best genotypes based on high breeding values for grain yield (kg/ha), we analyzed the historical data from the trials conducted in the IRRI, Philippines and Bangladesh. A two-stage mixed-model approach accounting for experimental design factors and a relationship matrix was fitted to obtain the breeding values for grain yield and estimate genetic trends. A positive genetic trend of 0.1% per annum with a yield advantage of 1.52 kg/ha was observed in IRRI, Philippines. In Bangladesh, we observed a genetic gain of 0.31% per annum with a yield advantage of 14.02 kg/ha. In the released varieties, we observed a genetic gain of 0.12% per annum with a 2.2 kg/ha/year yield advantage in the IRRI, Philippines. For the Bangladesh dataset, a genetic gain of 0.14% per annum with a yield advantage of 5.9 kg/ha/year was observed in the released varieties. Based on breeding values for grain yield, a core set of the top 145 genotypes with higher breeding values of > 2400 kg/ha in the IRRI, Philippines, and > 3500 kg/ha in Bangladesh with a reliability of > 0.4 were selected to develop the elite breeding panel. Conclusively, a recurrent selection breeding strategy integrated with novel technologies like genomic selection and speed breeding is highly required to achieve higher genetic gains in IRRI's salinity breeding programs.

Genetic architecture and genomic prediction of plant height-related traits in chrysanthemum

Plant height (PH) is a crucial trait determining plant architecture in chrysanthemum. To better understand the genetic basis of PH, we investigated the variations of PH, internode number (IN), internode length (IL), and stem diameter (SD) in a panel of 200 cut chrysanthemum accessions. Based on 330 710 high-quality SNPs generated by genotyping by sequencing, a total of 42 associations were identified via a genome-wide association study (GWAS), and 16 genomic regions covering 2.57 Mb of the whole genome were detected through selective sweep analysis. In addition, two SNPs, Chr1_339370594 and Chr18_230810045, respectively associated with PH and SD, overlapped with the selective sweep regions from FST and π ratios. Moreover, candidate genes involved in hormones, growth, transcriptional regulation, and metabolic processes were highlighted based on the annotation of homologous genes in Arabidopsis and transcriptomes in chrysanthemum. Finally, genomic selection for four PH-related traits was performed using a ridge regression best linear unbiased predictor model (rrBLUP) and six marker sets. The marker set constituting the top 1000 most significant SNPs identified via GWAS showed higher predictabilities for the four PH-related traits, ranging from 0.94 to 0.97. These findings improve our knowledge of the genetic basis of PH and provide valuable markers that could be applied in chrysanthemum genomic selection breeding programs.

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.

Challenges for crop improvement

The genetic improvement of crops faces the significant challenge of feeding an ever-increasing population amidst a changing climate, and when governments are adopting a 'more with less' approach to reduce input use. Plant breeding has the potential to contribute to the United Nations Agenda 2030 by addressing various sustainable development goals (SDGs), with its most profound impact expected on SDG2 Zero Hunger. To expedite the time-consuming crossbreeding process, a genomic-led approach for predicting breeding values, targeted mutagenesis through gene editing, high-throughput phenomics for trait evaluation, enviromics for including characterization of the testing environments, machine learning for effective management of large datasets, and speed breeding techniques promoting early flowering and seed production are being incorporated into the plant breeding toolbox. These advancements are poised to enhance genetic gains through selection in the cultigen pools of various crops. Consequently, these knowledge-based breeding methods are pursued for trait introgression, population improvement, and cultivar development. This article uses the potato crop as an example to showcase the progress being made in both genomic-led approaches and gene editing for accelerating the delivery of genetic gains through the utilization of genetically enhanced elite germplasm. It also further underscores that access to technological advances in plant breeding may be influenced by regulations and intellectual property rights.

Evaluation of eight Bayesian genomic prediction models for three micronutrient traits in bread wheat (Triticum aestivum L.)

In wheat, genomic prediction accuracy (GPA) was assessed for three micronutrient traits (grain iron, grain zinc, and β-carotenoid concentrations) using eight Bayesian regression models. For this purpose, data on 246 accessions, each genotyped with 17,937 DArT markers, were utilized. The phenotypic data on traits were available for 2013-2014 from Powerkheda (Madhya Pradesh) and for 2014-2015 from Meerut (Uttar Pradesh), India. The accuracy of the models was measured in terms of reliability, which was computed following a repeated cross-validation approach. The predictions were obtained independently for each of the two environments after adjusting for the local effects and across environments after adjusting for the environmental effects. The Bayes ridge regression (BayesRR) model outperformed the other seven models, whereas BayesLASSO (BayesL) was the least efficient. The GPA increased with an increase in the size of the training set as well as with an increase in marker density. The GPA values differed for the three traits and were higher for the best linear unbiased estimate (BLUE) (obtained after adjusting for the environmental effects) relative to those for the two environments. The GPA also remained unaffected after accounting for the population structure. The results of the present study suggest that only the best model should be used for the estimations of genomic estimated breeding values (GEBVs) before their use for genomic selection to improve the grain micronutrient contents.

Genetic interrogation of phenotypic plasticity informs genome-enabled breeding in cotton

Phenotypic plasticity, or the ability to adapt to and thrive in changing climates and variable environments, is essential for developmental programs in plants. Despite its importance, the genetic underpinnings of phenotypic plasticity for key agronomic traits remain poorly understood in many crops. In this study, we aim to fill this gap by using genome-wide association studies to identify genetic variations associated with phenotypic plasticity in upland cotton (Gossypium hirsutum L.). We identified 73 additive quantitative trait loci (QTLs), 32 dominant QTLs, and 6799 epistatic QTLs associated with 20 traits. We also identified 117 additive QTLs, 28 dominant QTLs, and 4691 epistatic QTLs associated with phenotypic plasticity in 19 traits. Our findings reveal new genetic factors, including additive, dominant, and epistatic QTLs, that are linked to phenotypic plasticity and agronomic traits. Meanwhile, we find that the genetic factors controlling the mean phenotype and phenotypic plasticity are largely independent in upland cotton, indicating the potential for simultaneous improvement. Additionally, we envision a genomic design strategy by utilizing the identified QTLs to facilitate cotton breeding. Taken together, our study provides new insights into the genetic basis of phenotypic plasticity in cotton, which should be valuable for future breeding.

Integrating genome-wide association study into genomic selection for the prediction of agronomic traits in rice (Oryza sativa L.)

Accurately identifying varieties with targeted agronomic traits was thought to contribute to genetic selection and accelerate rice breeding progress. Genomic selection (GS) is a promising technique that uses markers covering the whole genome to predict the genomic-estimated breeding values (GEBV), with the ability to select before phenotypes are measured. To choose the appropriate GS models for breeding work, we analyzed the predictability of nine agronomic traits measured from a population of 459 diverse rice varieties. By the comparison of eight representative GS models, we found that the prediction accuracies ranged from 0.407 to 0.896, with reproducing kernel Hilbert space (RKHS) having the highest predictive ability in most traits. Further results demonstrated the predictivity of GS is altered by several factors. Moreover, we assessed the method of integrating genome-wide association study (GWAS) into various GS models. The predictabilities of GS combined peak-associated markers generated from six different GWAS models were significantly different; a recommendation of Mixed Linear Model (MLM)-RKHS was given for the GWAS-GS-integrated prediction. Finally, based on the above result, we experimented with applying the P-values obtained from optimal GWAS models into ridge regression best linear unbiased prediction (rrBLUP), which benefited the low predictive traits in rice.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11032-023-01423-y.

SoyDNGP: a web-accessible deep learning framework for genomic prediction in soybean breeding

Soybean is a globally significant crop, playing a vital role in human nutrition and agriculture. Its complex genetic structure and wide trait variation, however, pose challenges for breeders and researchers aiming to optimize its yield and quality. Addressing this biological complexity requires innovative and accurate tools for trait prediction. In response to this challenge, we have developed SoyDNGP, a deep learning-based model that offers significant advancements in the field of soybean trait prediction. Compared to existing methods, such as DeepGS and DNNGP, SoyDNGP boasts a distinct advantage due to its minimal increase in parameter volume and superior predictive accuracy. Through rigorous performance comparison, including prediction accuracy and model complexity, SoyDNGP represents improved performance to its counterparts. Furthermore, it effectively predicted complex traits with remarkable precision, demonstrating robust performance across different sample sizes and trait complexities. We also tested the versatility of SoyDNGP across multiple crop species, including cotton, maize, rice and tomato. Our results showed its consistent and comparable performance, emphasizing SoyDNGP's potential as a versatile tool for genomic prediction across a broad range of crops. To enhance its accessibility to users without extensive programming experience, we designed a user-friendly web server, available at http://xtlab.hzau.edu.cn/SoyDNGP. The server provides two features: 'Trait Lookup', offering users the ability to access pre-existing trait predictions for over 500 soybean accessions, and 'Trait Prediction', allowing for the upload of VCF files for trait estimation. By providing a high-performing, accessible tool for trait prediction, SoyDNGP opens up new possibilities in the quest for optimized soybean breeding.

Phenomic data-driven biological prediction of maize through field-based high-throughput phenotyping integration with genomic data

High-throughput phenotyping (HTP) has expanded the dimensionality of data in plant research; however, HTP has resulted in few novel biological discoveries to date. Field-based HTP (FHTP), using small unoccupied aerial vehicles (UAVs) equipped with imaging sensors, can be deployed routinely to monitor segregating plant population interactions with the environment under biologically meaningful conditions. Here, flowering dates and plant height, important phenological fitness traits, were collected on 520 segregating maize recombinant inbred lines (RILs) in both irrigated and drought stress trials in 2018. Using UAV phenomic, single nucleotide polymorphism (SNP) genomic, as well as combined data, flowering times were predicted using several scenarios. Untested genotypes were predicted with 0.58, 0.59, and 0.41 prediction ability for anthesis, silking, and terminal plant height, respectively, using genomic data, but prediction ability increased to 0.77, 0.76, and 0.58 when phenomic and genomic data were used together. Using the phenomic data in a genome-wide association study, a heat-related candidate gene (GRMZM2G083810; hsp18f) was discovered using temporal reflectance phenotypes belonging to flowering times (both irrigated and drought) trials where heat stress also peaked. Thus, a relationship between plants and abiotic stresses belonging to a specific time of growth was revealed only through use of temporal phenomic data. Overall, this study showed that (i) it is possible to predict complex traits using high dimensional phenomic data between different environments, and (ii) temporal phenomic data can reveal a time-dependent association between genotypes and abiotic stresses, which can help understand mechanisms to develop resilient plants.

The pursuit of genetic gain in agricultural crops through the application of machine-learning to genomic prediction

Current practice in agriculture applies genomic prediction to assist crop breeding in the analysis of genetic marker data. Genomic selection methods typically use linear mixed models, but using machine-learning may provide further potential for improved selection accuracy, or may provide additional information. Here we describe SelectML, an automated pipeline for testing and comparing the performance of a range of linear mixed model and machine-learning-based genomic selection methods. We demonstrate the use of SelectML on an in silico-generated marker dataset which simulated a randomly-sampled (mixed) and an unevenly-sampled (unbalanced) population, comparing the relative performance of various methods included in SelectML on the two datasets. Although machine-learning based methods performed similarly overall to linear mixed models, they performed worse on the mixed dataset and marginally better on the unbalanced dataset, being more affected than linear mixed models by the imposed sampling bias. SelectML can assist in the training, comparison, and selection of genomic selection models, and is available from https://github.com/darcyabjones/selectml.

The allotetraploid horseradish genome provides insights into subgenome diversification and formation of critical traits

Polyploidization can provide a wealth of genetic variation for adaptive evolution and speciation, but understanding the mechanisms of subgenome evolution as well as its dynamics and ultimate consequences remains elusive. Here, we report the telomere-to-telomere (T2T) gap-free reference genome of allotetraploid horseradish (Armoracia rusticana) sequenced using a comprehensive strategy. The (epi)genomic architecture and 3D chromatin structure of the A and B subgenomes differ significantly, suggesting that both the dynamics of the dominant long terminal repeat retrotransposons and DNA methylation have played critical roles in subgenome diversification. Investigation of the genetic basis of biosynthesis of glucosinolates (GSLs) and horseradish peroxidases reveals both the important role of polyploidization and subgenome differentiation in shaping the key traits. Continuous duplication and divergence of essential genes of GSL biosynthesis (e.g., FMO_GS-OX, IGMT, and GH1 gene family) contribute to the broad GSL profile in horseradish. Overall, the T2T assembly of the allotetraploid horseradish genome expands our understanding of polyploid genome evolution and provides a fundamental genetic resource for breeding and genetic improvement of horseradish.

Do feature selection methods for selecting environmental covariables enhance genomic prediction accuracy?

Genomic selection (GS) is transforming plant and animal breeding, but its practical implementation for complex traits and multi-environmental trials remains challenging. To address this issue, this study investigates the integration of environmental information with genotypic information in GS. The study proposes the use of two feature selection methods (Pearson's correlation and Boruta) for the integration of environmental information. Results indicate that the simple incorporation of environmental covariates may increase or decrease prediction accuracy depending on the case. However, optimal incorporation of environmental covariates using feature selection significantly improves prediction accuracy in four out of six datasets between 14.25% and 218.71% under a leave one environment out cross validation scenario in terms of Normalized Root Mean Squared Error, but not relevant gain was observed in terms of Pearson´s correlation. In two datasets where environmental covariates are unrelated to the response variable, feature selection is unable to enhance prediction accuracy. Therefore, the study provides empirical evidence supporting the use of feature selection to improve the prediction power of GS.

The Genetic Structures and Molecular Mechanisms Underlying Ear Traits in Maize (Zea mays L.)

Maize (Zea mays L.) is one of the world's staple food crops. In order to feed the growing world population, improving maize yield is a top priority for breeding programs. Ear traits are important determinants of maize yield, and are mostly quantitatively inherited. To date, many studies relating to the genetic and molecular dissection of ear traits have been performed; therefore, we explored the genetic loci of the ear traits that were previously discovered in the genome-wide association study (GWAS) and quantitative trait locus (QTL) mapping studies, and refined 153 QTL and 85 quantitative trait nucleotide (QTN) clusters. Next, we shortlisted 19 common intervals (CIs) that can be detected simultaneously by both QTL mapping and GWAS, and 40 CIs that have pleiotropic effects on ear traits. Further, we predicted the best possible candidate genes from 71 QTL and 25 QTN clusters that could be valuable for maize yield improvement.

Integrated Genomic Selection for Accelerating Breeding Programs of Climate-Smart Cereals

Rapidly rising population and climate changes are two critical issues that require immediate action to achieve sustainable development goals. The rising population is posing increased demand for food, thereby pushing for an acceleration in agricultural production. Furthermore, increased anthropogenic activities have resulted in environmental pollution such as water pollution and soil degradation as well as alterations in the composition and concentration of environmental gases. These changes are affecting not only biodiversity loss but also affecting the physio-biochemical processes of crop plants, resulting in a stress-induced decline in crop yield. To overcome such problems and ensure the supply of food material, consistent efforts are being made to develop strategies and techniques to increase crop yield and to enhance tolerance toward climate-induced stress. Plant breeding evolved after domestication and initially remained dependent on phenotype-based selection for crop improvement. But it has grown through cytological and biochemical methods, and the newer contemporary methods are based on DNA-marker-based strategies that help in the selection of agronomically useful traits. These are now supported by high-end molecular biology tools like PCR, high-throughput genotyping and phenotyping, data from crop morpho-physiology, statistical tools, bioinformatics, and machine learning. After establishing its worth in animal breeding, genomic selection (GS), an improved variant of marker-assisted selection (MAS), has made its way into crop-breeding programs as a powerful selection tool. To develop novel breeding programs as well as innovative marker-based models for genetic evaluation, GS makes use of molecular genetic markers. GS can amend complex traits like yield as well as shorten the breeding period, making it advantageous over pedigree breeding and marker-assisted selection (MAS). It reduces the time and resources that are required for plant breeding while allowing for an increased genetic gain of complex attributes. It has been taken to new heights by integrating innovative and advanced technologies such as speed breeding, machine learning, and environmental/weather data to further harness the GS potential, an approach known as integrated genomic selection (IGS). This review highlights the IGS strategies, procedures, integrated approaches, and associated emerging issues, with a special emphasis on cereal crops. In this domain, efforts have been taken to highlight the potential of this cutting-edge innovation to develop climate-smart crops that can endure abiotic stresses with the motive of keeping production and quality at par with the global food demand.

Next-Generation-Sequencing-Based Simple Sequence Repeat (SSR) Marker Development and Linkage Mapping in Lentil (Lens culinaris L.)

Simple sequence repeats (SSRs) are highly versatile markers in genetic diversity analysis and plant breeding, making them widely applicable. They hold potential in lentil (Lens culinaris) breeding for genetic diversity analysis, marker-assisted selection (MAS), and linkage mapping. However, the availability and diversity of SSR markers in lentil is limited. We used next-generation sequencing (NGS) technology to develop SSR markers in lentil. NGS allowed us to identify regions of the lentil genome that contained SSRs. Illumina Hiseq-2000 sequencing of the lentil genotype "Karacadağ" resulted in 1,727,734 sequence reads comprising more than 48,390 Mb, and contigs were mined for SSRs, resulting in the identification of a total of 8697 SSR motifs. Among these, dinucleotide repeats were the most abundant (53.38%), followed by trinucleotides (30.38%), hexanucleotides (6.96%), tetranucleotides (6.59%), and pentanucleotides (3.19%). The most frequent repeat in dinucleotides was the TC (21.80%), followed by the GA (17.60%). A total of 2000 primer pairs were designed from these motifs, and 458 SSR markers were validated following their amplified PCR products. A linkage map was constructed using these new SSRs with high linkage disequilibrium (209) and previously known SSRs (11). The highest number of SSR markers (43) was obtained in LG2, while the lowest number of SSR markers (19) was obtained in LG7. The longest linkage group (LG) was LG2 (86.84 cM), whereas the shortest linkage group was LG7 (53.46 cM). The average length between markers ranged from 1.86 cM in LG1 to 2.81 cM in LG7, and the map density was 2.16 cM. The developed SSRs and created linkage map may provide useful information and offer a new library for genetic diversity analyses, linkage mapping studies, and lentil breeding programs.

Partial least squares enhance multi-trait genomic prediction of potato cultivars in new environments

It is of paramount importance in plant breeding to have methods dealing with large numbers of predictor variables and few sample observations, as well as efficient methods for dealing with high correlation in predictors and measured traits. This paper explores in terms of prediction performance the partial least squares (PLS) method under single-trait (ST) and multi-trait (MT) prediction of potato traits. The first prediction was for tested lines in tested environments under a five-fold cross-validation (5FCV) strategy and the second prediction was for tested lines in untested environments (herein denoted as leave one environment out cross validation, LOEO). There was a good performance in terms of predictions (with accuracy mostly > 0.5 for Pearson's correlation) the accuracy of 5FCV was better than LOEO. Hence, we have empirical evidence that the ST and MT PLS framework is a very valuable tool for prediction in the context of potato breeding data.

Optimizing genomic selection of agricultural traits using K-wheat core collection

The agricultural traits that constitute basic plant breeding information are usually quantitative or complex in nature. This quantitative and complex combination of traits complicates the process of selection in breeding. This study examined the potential of genome-wide association studies (GWAS) and genomewide selection (GS) for breeding ten agricultural traits by using genome-wide SNPs. As a first step, a trait-associated candidate marker was identified by GWAS using a genetically diverse 567 Korean (K)-wheat core collection. The accessions were genotyped using an Axiom^® 35K wheat DNA chip, and ten agricultural traits were determined (awn color, awn length, culm color, culm length, ear color, ear length, days to heading, days to maturity, leaf length, and leaf width). It is essential to sustain global wheat production by utilizing accessions in wheat breeding. Among the traits associated with awn color and ear color that showed a high positive correlation, a SNP located on chr1B was significantly associated with both traits. Next, GS evaluated the prediction accuracy using six predictive models (G-BLUP, LASSO, BayseA, reproducing kernel Hilbert space, support vector machine (SVM), and random forest) and various training populations (TPs). With the exception of the SVM, all statistical models demonstrated a prediction accuracy of 0.4 or better. For the optimization of the TP, the number of TPs was randomly selected (10%, 30%, 50% and 70%) or divided into three subgroups (CC-sub 1, CC-sub 2 and CC-sub 3) based on the subpopulation structure. Based on subgroup-based TPs, better prediction accuracy was found for awn color, culm color, culm length, ear color, ear length, and leaf width. A variety of Korean wheat cultivars were used for validation to evaluate the prediction ability of populations. Seven out of ten cultivars showed phenotype-consistent results based on genomics-evaluated breeding values (GEBVs) calculated by the reproducing kernel Hilbert space (RKHS) predictive model. Our research provides a basis for improving complex traits in wheat breeding programs through genomics assisted breeding. The results of our research can be used as a basis for improving wheat breeding programs by using genomics-assisted breeding.

Underutilized legumes: nutrient status and advanced breeding approaches for qualitative and quantitative enhancement

Underutilized/orphan legumes provide food and nutritional security to resource-poor rural populations during periods of drought and extreme hunger, thus, saving millions of lives. The Leguminaceae, which is the third largest flowering plant family, has approximately 650 genera and 20,000 species and are distributed globally. There are various protein-rich accessible and edible legumes, such as soybean, cowpea, and others; nevertheless, their consumption rate is far higher than production, owing to ever-increasing demand. The growing global urge to switch from an animal-based protein diet to a vegetarian-based protein diet has also accelerated their demand. In this context, underutilized legumes offer significant potential for food security, nutritional requirements, and agricultural development. Many of the known legumes like Mucuna spp., Canavalia spp., Sesbania spp., Phaseolus spp., and others are reported to contain comparable amounts of protein, essential amino acids, polyunsaturated fatty acids (PUFAs), dietary fiber, essential minerals and vitamins along with other bioactive compounds. Keeping this in mind, the current review focuses on the potential of discovering underutilized legumes as a source of food, feed and pharmaceutically valuable chemicals, in order to provide baseline data for addressing malnutrition-related problems and sustaining pulse needs across the globe. There is a scarcity of information about underutilized legumes and is restricted to specific geographical zones with local or traditional significance. Around 700 genera and 20,000 species remain for domestication, improvement, and mainstreaming. Significant efforts in research, breeding, and development are required to transform existing local landraces of carefully selected, promising crops into types with broad adaptability and economic viability. Different breeding efforts and the use of biotechnological methods such as micro-propagation, molecular markers research and genetic transformation for the development of underutilized crops are offered to popularize lesser-known legume crops and help farmers diversify their agricultural systems and boost their profitability.

Combining phenotypic and genomic data to improve prediction of binary traits

Plant breeders want to develop cultivars that outperform existing genotypes. Some characteristics (here 'main traits') of these cultivars are categorical and difficult to measure directly. It is important to predict the main trait of newly developed genotypes accurately. In addition to marker data, breeding programs often have information on secondary traits (or 'phenotypes') that are easy to measure. Our goal is to improve prediction of main traits with interpretable relations by combining the two data types using variable selection techniques. However, the genomic characteristics can overwhelm the set of secondary traits, so a standard technique may fail to select any phenotypic variables. We develop a new statistical technique that ensures appropriate representation from both the secondary traits and the genotypic variables for optimal prediction. When two data types (markers and secondary traits) are available, we achieve improved prediction of a binary trait by two steps that are designed to ensure that a significant intrinsic effect of a phenotype is incorporated in the relation before accounting for extra effects of genotypes. First, we sparsely regress the secondary traits on the markers and replace the secondary traits by their residuals to obtain the effects of phenotypic variables as adjusted by the genotypic variables. Then, we develop a sparse logistic classifier using the markers and residuals so that the adjusted phenotypes may be selected first to avoid being overwhelmed by the genotypic variables due to their numerical advantage. This classifier uses forward selection aided by a penalty term and can be computed effectively by a technique called the one-pass method. It compares favorably with other classifiers on simulated and real data.

Optimal implementation of genomic selection in clone breeding programs-Exemplified in potato: I. Effect of selection strategy, implementation stage, and selection intensity on short-term genetic gain

Genomic selection (GS) is used in many animal and plant breeding programs to enhance genetic gain for complex traits. However, its optimal integration in clone breeding programs, for example potato, that up to now relied on phenotypic selection (PS) requires further research. In this study, we performed computer simulations based on an empirical genomic dataset of tetraploid potato to (i) investigate under a fixed budget how the weight of GS relative to PS, the stage of implementing GS, the correlation between an auxiliary trait and the target trait, the variance components, and the prediction accuracy affect the genetic gain of the target trait, (ii) determine the optimal allocation of resources maximizing the genetic gain of the target trait, and (iii) make recommendations to breeders how to implement GS in clone and especially potato breeding programs. In our simulation results, any selection strategy involving GS had a higher short-term genetic gain for the target trait than Standard-PS. In addition, we showed that implementing GS in consecutive selection stages can largely enhance short-term genetic gain and recommend the breeders to implement GS at single hills and A clone stages. Furthermore, we observed for selection strategies involving GS that the optimal allocation of resources maximizing the genetic gain of the target trait differed considerably from those typically used in potato breeding programs and, thus, require the adjustment of the selection and phenotyping intensities. The trends are described in our study. Therefore, our study provides new insight for breeders regarding how to optimally implement GS in a commercial potato breeding program to improve the short-term genetic gain for their target trait.

Understanding the genomic selection for crop improvement: current progress and future prospects

Although increased use of modern breeding techniques and technology has resulted in long-term genetic gain, the pace of genetic gain must be sped up to satisfy global agricultural demand. However, marker-assisted selection has proven its potential for improving qualitative traits with large effects regulated by one to few genes. Its contribution to the improvement of the quantitative traits regulated by a number of small-effect genes is modest. In this context, genomic selection (GS) has been regarded as the most promising method for genetically enhancing complicated features that are regulated by several genes, each of which has minor effects. By examining a population's phenotypes and high-density marker scores, genomic selection can forecast the breeding potential of individual lines. The fact that GS uses all marker data in the prediction model prevents skewed marker effect estimations and maximizes the amount of variation caused by small-effect QTL. It has the ability to speed up the breeding cycle and as a consequence of which superior genotypes are selected rapidly. Developing the best GS models while taking into account non-additive effects, genotype-by-environment interaction, and cost-effectiveness will enable the widespread implementation of GS in plants. These steps will also increase heritability estimation and prediction accuracy. This review focuses on the shift from conventional selection methods to GS, underlying statistical tools and methodologies, the state of GS research in agricultural plants, and prospects for its effective use in the creation of climate-resilient crops.

Efficacy of plant breeding using genomic information

Genomic selection (GS) proposed by Meuwissen et al. more than 20 years ago, is revolutionizing plant and animal breeding. Although GS has been widely accepted and applied to plant and animal breeding, there are many factors affecting its efficacy. We studied 14 real datasets to respond to the practical question of whether the accuracy of genomic prediction increases when considering genomic as compared with not using genomic. We found across traits, environments, datasets, and metrics, that the average gain in prediction accuracy when genomic information is considered was 26.31%, while only in terms of Pearson's correlation the gain was of 46.1%, while only in terms of normalized root mean squared error the gain was of 6.6%. If the quality of the makers and relatedness of the individuals increase, major gains in prediction accuracy can be obtained, but if these two factors decrease, a lower increase is possible. Finally, our findings reinforce genomic is vital for improving the prediction accuracy and, therefore, the realized genetic gain in genomic assisted plant breeding programs.

Two simple methods to improve the accuracy of the genomic selection methodology

BACKGROUND

Genomic selection (GS) is revolutionizing plant and animal breeding. However, still its practical implementation is challenging since it is affected by many factors that when they are not under control make this methodology not effective. Also, due to the fact that it is formulated as a regression problem in general has low sensitivity to select the best candidate individuals since a top percentage is selected according to a ranking of predicted breeding values.

RESULTS

For this reason, in this paper we propose two methods to improve the prediction accuracy of this methodology. One of the methods consist in reformulating the GS (nowadays formulated as a regression problem) methodology as a binary classification problem. The other consists only in a postprocessing step that adjust the threshold used for classification of the lines predicted in its original scale (continues scale) to guarantee similar sensitivity and specificity. The postprocessing method is applied for the resulting predictions after obtaining the predictions using the conventional regression model. Both methods assume that we defined with anticipation a threshold, to divide the training data as top lines and not top lines, and this threshold can be decided in terms of a quantile (for example 80%, 90%, etc.) or as the average (or maximum) of the performance of the checks. In the reformulation method it is required to label as one those lines in the training set that are equal or larger than the specified threshold and as zero otherwise. Then we train a binary classification model with the conventional inputs, but using the binary response variable in place of the continuous response variable. The training of the binary classification should be done to guarantee a more similar sensitivity and specificity, to guarantee a reasonable probability of classification of the top lines.

CONCLUSIONS

We evaluated the proposed models in seven data sets and we found that the two proposed methods outperformed by large margin the conventional regression model (by 402.9% in terms of sensitivity, by 110.04% in terms of F1 score and by 70.96% in terms of Kappa coefficient, with the postprocessing methods). However, between the two proposed methods the postprocessing method was better than the reformulation as binary classification model. The simple postprocessing method to improve the accuracy of the conventional genomic regression models avoid the need to reformulate the conventional regression models as binary classification models with similar or better performance, that significantly improve the selection of the top best candidate lines. In general both proposed methods are simple and can easily be adopted for use in practical breeding programs, with the guarantee that will improve significantly the selection of the top best candidates lines.

Haplotype-tagged SNPs improve genomic prediction accuracy for Fusarium head blight resistance and yield-related traits in wheat

Linkage disequilibrium (LD)-based haplotyping with subsequent SNP tagging improved the genomic prediction accuracy up to 0.07 and 0.092 for Fusarium head blight resistance and spike width, respectively, across six different models. Genomic prediction is a powerful tool to enhance genetic gain in plant breeding. However, the method is accompanied by various complications leading to low prediction accuracy. One of the major challenges arises from the complex dimensionality of marker data. To overcome this issue, we applied two pre-selection methods for SNP markers viz. LD-based haplotype-tagging and GWAS-based trait-linked marker identification. Six different models were tested with preselected SNPs to predict the genomic estimated breeding values (GEBVs) of four traits measured in 419 winter wheat genotypes. Ten different sets of haplotype-tagged SNPs were selected by adjusting the level of LD thresholds. In addition, various sets of trait-linked SNPs were identified with different scenarios from the training-test combined and only from the training populations. The BRR and RR-BLUP models developed from haplotype-tagged SNPs had a higher prediction accuracy for FHB and SPW by 0.07 and 0.092, respectively, compared to the corresponding models developed without marker pre-selection. The highest prediction accuracy for SPW and FHB was achieved with tagged SNPs pruned at weak LD thresholds (r² < 0.5), while stringent LD was required for spike length (SPL) and flag leaf area (FLA). Trait-linked SNPs identified only from training populations failed to improve the prediction accuracy of the four studied traits. Pre-selection of SNPs via LD-based haplotype-tagging could play a vital role in optimizing genomic selection and reducing genotyping costs. Furthermore, the method could pave the way for developing low-cost genotyping methods through customized genotyping platforms targeting key SNP markers tagged to essential haplotype blocks.

IP4GS: Bringing genomic selection analysis to breeders

Genomic selection (GS), a strategy to use genotypes to predict phenotypes via statistical or machine learning models, has become a routine practice in plant breeding programs. GS can speed up the genetic gain by reducing phenotyping costs and/or shortening the breeding cycles. GS analysis is complicated involving data clean up and formatting, training and test population analysis, model selection and evaluation, and parameter optimization. In addition, GS analysis also requires some programming skills and knowledge of statistical modeling. Thus, we need a more practical GS tools for breeders. To alleviate this difficulty, we developed the web-based platform IP4GS (https://ngdc.cncb.ac.cn/ip4gs/), which offers a user-friendly interface to perform GS analysis simply through point-and-click actions. IP4GS currently includes seven commonly used models, eleven evaluation metrics, and visualization modules, offering great convenience for plant breeders with limited bioinformatics knowledge to apply GS analysis.

High-precision early warning system for rice cadmium accumulation risk assessment

Rapid global industrialization has resulted in widespread cadmium contamination in agricultural soils and products. A considerable proportion of rice consumers are exposed to Cd levels above the provisional safe intake limit, raising widespread environmental concerns on risk management. Therefore, a generalized approach is urgently needed to enable correct evaluation and early warning of cadmium contaminants in rice products. Combining big data and computer science together, this study developed a system named "SMART Cd Early Warning", which integrated 4 modules including genotype-to-phenotype (G2P) modelling, high-throughput sequencing, G2P prediction and rice Cd contamination risk assessment, for rice cadmium accumulation early warning. This system can rapidly assess the risk of rice cadmium accumulation by genotyping leaves at seeding stage. The parameters including statistical methods, population size, training population-testing population ratio, SNP density were assessed to ensure G2P model exhibited superior performance in terms of prediction precision (up to 0.76 ± 0.003) and computing efficiency (within 2 h). In field trials of cadmium-contaminated farmlands in Wenling and Fuyang city, Zhejiang Province, "SMART Cd Early Warning" exhibited superior capability for identification risk rice varieties, suggesting a potential of "SMART Cd Early-Warning system" in OsGCd risk assessment and early warning in the age of smart.

Integration of high-throughput phenotyping, GWAS, and predictive models reveals the genetic architecture of plant height in maize

Plant height (PH) is an essential trait in maize (Zea mays) that is tightly associated with planting density, biomass, lodging resistance, and grain yield in the field. Dissecting the dynamics of maize plant architecture will be beneficial for ideotype-based maize breeding and prediction, as the genetic basis controlling PH in maize remains largely unknown. In this study, we developed an automated high-throughput phenotyping platform (HTP) to systematically and noninvasively quantify 77 image-based traits (i-traits) and 20 field traits (f-traits) for 228 maize inbred lines across all developmental stages. Time-resolved i-traits with novel digital phenotypes and complex correlations with agronomic traits were characterized to reveal the dynamics of maize growth. An i-trait-based genome-wide association study identified 4945 trait-associated SNPs, 2603 genetic loci, and 1974 corresponding candidate genes. We found that rapid growth of maize plants occurs mainly at two developmental stages, stage 2 (S2) to S3 and S5 to S6, accounting for the final PH indicators. By integrating the PH-association network with the transcriptome profiles of specific internodes, we revealed 13 hub genes that may play vital roles during rapid growth. The candidate genes and novel i-traits identified at multiple growth stages may be used as potential indicators for final PH in maize. One candidate gene, ZmVATE, was functionally validated and shown to regulate PH-related traits in maize using genetic mutation. Furthermore, machine learning was used to build predictive models for final PH based on i-traits, and their performance was assessed across developmental stages. Moderate, strong, and very strong correlations between predictions and experimental datasets were achieved from the early S4 (tenth-leaf) stage. Colletively, our study provides a valuable tool for dissecting the spatiotemporal formation of specific internodes and the genetic architecture of PH, as well as resources and predictive models that are useful for molecular design breeding and predicting maize varieties with ideal plant architectures.

Integrating Parental Phenotypic Data Enhances Prediction Accuracy of Hybrids in Wheat Traits

Genomic selection (GS) is a methodology that is revolutionizing plant breeding because it can select candidate genotypes without phenotypic evaluation in the field. However, its practical implementation in hybrid prediction remains challenging since many factors affect its accuracy. The main objective of this study was to research the genomic prediction accuracy of wheat hybrids by adding covariates with the hybrid parental phenotypic information to the model. Four types of different models (MA, MB, MC, and MD) with one covariate (same trait to be predicted) (MA_C, MB_C, MC_C, and MD_C) or several covariates (of the same trait and other correlated traits) (MA_AC, MB_AC, MC_AC, and MD_AC) were studied. We found that the four models with parental information outperformed models without parental information in terms of mean square error by at least 14.1% (MA vs. MA_C), 5.5% (MB vs. MB_C), 51.4% (MC vs. MC_C), and 6.4% (MD vs. MD_C) when parental information of the same trait was used and by at least 13.7% (MA vs. MA_AC), 5.3% (MB vs. MB_AC), 55.1% (MC vs. MC_AC), and 6.0% (MD vs. MD_AC) when parental information of the same trait and other correlated traits were used. Our results also show a large gain in prediction accuracy when covariates were considered using the parental phenotypic information, as opposed to marker information. Finally, our results empirically demonstrate that a significant improvement in prediction accuracy was gained by adding parental phenotypic information as covariates; however, this is expensive since, in many breeding programs, the parental phenotypic information is unavailable.