Improving Genomic Prediction Using High-Dimensional Secondary Phenotypes: The Genetic Latent Factor Approach.

Multiple environment phenotypes may be utilized to implement genomic prediction in plant breeding, while it is unclear about optimal utilization strategies according to its different availability. It is necessary to assess the utilization strategies of genomic prediction models based on different availability of multiple environment phenotypes. Here, we compared the prediction accuracy of three genomic prediction models (genomic prediction model (genomic best linear unbiased prediction (GBLUP), genomic best linear unbiased prediction (GFBLUP), and multi-trait genomic best linear unbiased prediction (mtGBLUP)) which leveraged diverse information from multiple environment phenotypes using a rice dataset containing 19 agronomic traits in two disparate seasons. We found that the prediction accuracy of genomic prediction models considering multiple environment phenotypes (GFBLUP and mtGBLUP) was better than the classical genomic prediction model (GBLUP model). The deviation of prediction accuracy of between GBLUP and mtGBLUP or GFBLUP was associated with the phenotypic correlation. In summary, the genomic prediction models considering multiple environment phenotypes (GFBLUP and mtGBLUP) demonstrated better prediction accuracy. In addition, we could utilize different genomic prediction strategies according to different availability of multiple environment phenotypes.

Many quantitative traits measured in breeding programs are genetically correlated. The genetic correlations between the traits indicate that the measurement of one trait carries information on others. To benefit from this information, multi-trait genomic prediction (MTGP) is preferable to use. However, MTGP is more difficult to implement compared to single-trait genomic prediction (STGP), and even more challenging for the goal to exploit not only the information on other traits but also the information on ungenotyped animals. This could be accomplished using both single and multistep methods. The single-step method was achieved by implementing a single-step genomic best linear unbiased prediction (ssGBLUP) approach using a multi-trait model. Here, we examined a multistep analysis based on an approach called "Absorption" to achieve this goal. The Absorption approach absorbed all available information including the phenotypic information on ungenotyped animals as well as the information on other traits if applicable, into mixed model equations of genotyped animals. The multistep analysis included (1) to apply the Absorption approach that exploits all available information and (2) to implement genomic BLUP (GBLUP) prediction on the absorbed dataset. In this study, the ssGBLUP and multistep analysis were applied to 5 traits in Duroc pigs, which were slaughter percentage, feed consumption from 40 to 120 kg (FC40_120), days of growth from 40 to 120 kg (D40_120), age at 40 kg (A40) and lean meat percentage. The results showed that MTGP yielded higher accuracy than STGP, which on average was 0.057 higher for the multistep method and 0.045 higher for ssGBLUP. The multistep method achieved similar prediction accuracy as ssGBLUP. However, the prediction bias of the multistep method was in general lower than that of ssGBLUP.

Genomic prediction leverages all available markers, irrespective of their statistical significance in genome-wide association studies (GWAS). Recent advancements in marker density, sample sizes, and sophisticated statistical GWAS methods have demonstrated that integrating GWAS results can potentially boost the accuracy of genomic predictions. The Genomic Association and Prediction Tool (GAPIT) has recently begun incorporating GWAS findings into its prediction framework, streamlining this approach, referred to as GWAS-Assisted Genomic Best Linear Unbiased Prediction (GAGBLUP). A sufficient simulation study revealed that the benefits of GAGBLUP depend on the GWAS model used. Multiple-locus models, such as Bayesian information and linkage-disequilibrium iteratively nested keyway (BLINK), outperformed single-locus models, such as the mixed linear model. Specifically, when BLINK GWAS results in a real trait were incorporated into Genomic Best Linear Unbiased Prediction (GBLUP), prediction accuracy improved by over 20% compared to GBLUP alone. This approach integrates the trait-specific insights from GWAS with the polygenic modeling capacity of GBLUP, resulting in more stable prediction across varying genetic backgrounds. This broader applicability enhances the utility of genomic selection in breeding programs, enabling its deployment across a wider range of crops and trait architectures.

BackgroundGene expression profiles hold potentially valuable information for the prediction of breeding values and phenotypes. However, in practical breeding programs, most reference population individuals typically have only genomic data, lacking transcriptomic data. Predicting gene expression based on genetic markers and integrating the genetically predicted gene expression data into genomic prediction may offer a potential solution.ResultsThis study extends kernel ridge regression (KRR) to weighted multiple kernel ridge regression (WMKRR), which integrates genomic data and transcriptomic data predicted from genetic markers through a multiple kernel learning (MKL) approach. We evaluated the predictive ability of WMKRR compared to traditional genomic best linear unbiased prediction (GBLUP) and a combined genomic and transcriptomic best linear unbiased prediction (GTBLUP) in both genotype feature selection and non-feature selection scenarios in two datasets: (i) 3305 simulated data based on the Cattle Genotype-Tissue Expression (CattleGTEx) dataset, (ii) 5515 real dairy cattle data. Our results show that WMKRR yielded higher predictive abilities than GBLUP And GTBLUP in both simulated And real dairy cattle data. For the simulated data based on CattleGTEx, WMKRR achieved an average improvement in predictive ability of 1.12% And 1.13% over GBLUP And GTBLUP, respectively, under the non-feature selection scenario, And 3.17% And 3.23%, respectively, under the feature selection scenario. For the real dairy cattle data, in cross-validation, WMKRR improved over GBLUP And GTBLUP by An average of 5.56% And 7.23%, respectively, without feature selection, And by 5.66% And 6.40%, respectively, with feature selection. In forward validation, WMKRR improved over GBLUP And GTBLUP by An average of 5.68% And 8.41%, respectively, without feature selection, And by 4.66% And 7.06%, respectively, with feature selection.ConclusionsOur result demonstrates that the WMKRR model, which integrates genomic and genetically predicted transcriptomic data, achieves better prediction performance compared to traditional genomic prediction models. This study showed the potential of enhanced genomic breeding application using omics data with no further omics sequencing cost.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12711-025-00997-9.

Natural or artificial selection could shape genetic architecture, e.g., the relationship between minor allele frequency (MAF) and the effect sizes of causal variants (CVs). This study aimed to investigate the impact of the MAF-effect size relationship (as a selection signature, S) on genomic prediction and heritability estimation in livestock, using both simulated data (Holstein) and real datasets (Holstein and pigs). We evaluated the performance of two models: (1) selection-adjusted genomic best linear unbiased prediction (GBLUP-S), and (2) MAF-stratified selection-adjusted genomic best linear unbiased prediction (GBLUP-SMS). Simulation results demonstrated that for traits under strong negative selection (S < -1), both GBLUP-S and GBLUP-SMS outperformed classic GBLUP. The prediction accuracy of GBLUP-S improved by 0.011-0.031, while GBLUP-SMS achieved a gain of 0.005-0.025. Furthermore, GBLUP-SMS exhibited lower sensitivity to variations in S-values, whereas GBLUP-S heavily relied on accurate S specification. When the true S was matched, GBLUP-SMS generated more unbiased (or comparable) heritability estimates and higher prediction accuracy relative to GBLUP-S. Critically, mismatched S in GBLUP-S led to increased bias in heritability estimates and reduced prediction accuracy. Cross-validation with real phenotypic data from Holsteins and pigs demonstrated that implementing selection-adjust methods improved prediction accuracy by 0.015 for FP in Holsteins and 0.01 for T1 in pigs, while enhancing the unbiasedness of heritability estimates across all traits. Negative selection signatures were identified for cattle (S = -0.5) and pig T1, T2, and T3 (S = -1.5, -1, and -2, respectively). These findings advance the theoretical framework of GBLUP-based genomic prediction and heritability estimation.

The detection and characterization of physiological processes in crop plants under water-limited conditions is essential for the selection of drought-tolerant genotypes and the functional analysis of related genes. Close-range hyperspectral imaging (HSI) is a promising, non-invasive technique for sensing of plant traits, and has the potential to detect plant responses to water deficit stress at an early stage. The present study describes a data analysis method to realize this potential. Reflectance spectra of plants in close-range imaging are highly influenced by illumination effects. Standard normal variate (SNV) was applied to reduce linear illumination effects, while non-linear effects were filtered by discarding the affected pixels through a clustering procedure. Once the illumination effects were eliminated, the remaining differences in plant spectra were assumed to be related to changes in plant traits. To quantify stress-related spectral dynamics, a spectral analysis procedure was developed based on a supervised band selection and a direct calculation of a spectral similarity measure against a reference. The proposed method was tested on HSI data of maize plants acquired in a high-throughput plant phenotyping platform for assessment of drought stress responses and recovery after re-watering events. Results show that the spectral analysis method successfully detected the drought stress responses at an early stage and consistently revealed the recovery effects shortly after the re-watering period.

Biological control is a sustainable strategy to combat agricultural pests. Yet, legislation increasingly restricts importing nonnative biocontrol agents. Thus, selective breeding of biocontrol traits is suggested to enhance performance of existing biocontrol agents. Genomic prediction, where genomic data are used to estimate the genetic merit of an individual for specific traits, is an alternative to exploit genetic variation for the improvement of native biocontrol agents. This study aims to establish a proof of principle for genomic prediction in insect biocontrol agents, using wing morphology traits in the model parasitoid Nasonia vitripennis Walker (Pteromalidae). We performed genomic prediction using a genomic best linear unbiased prediction (GBLUP) model, using 1,230 individuals with 8,639 SNPs generated by genotyping-by-sequencing (GBS). We used individuals from 2 generations from the outbred HVRx population, 717 individuals from generation 169 (G169) and 513 individuals from generation 172 (G172). To assess genomic prediction accuracy, we used across generation validation (forward validation for G172 from G169 and backward validation for G169 from G172) and also 5-fold cross-validation. For size-related traits, including tibia length, wing length, wing width, and second moment area, the accuracy of genomic prediction was close to 0 in both across generation validations but much higher in 5-fold cross-validation (ranging from 0.54 to 0.68). For the shape-related trait wing aspect ratio, a high accuracy was found for all 3 validation strategies, with 0.47 for across generation forward validation (AGFV), 0.65 for across generation backward validation (AGBV), and 0.54 for 5-fold cross-validation. Overall, genomic selection in insect biocontrol agents with a relative small effective population size seems promising. However, factors such as the biology of insects, phenotyping techniques, and large-scale genotyping costs still challenge the application of genomic selection to biocontrol agents.

Catfish is an important aquaculture species in the USA. Columnaris disease is distributed worldwide, affecting a wide variety of fish species including catfish . It leads to huge economic losses each year to the US catfish industry. Channel catfish in general is highly resistant to the disease, while blue catfish is highly susceptible. Genomic selection is an effective and accurate way to predict the breeding values and thus was expected to improve the prediction veracity of columnaris disease resistance in catfish effectively. In this study, two different methods, elastic net genomic best linear unbiased prediction (ENGBLUP) and genomic best linear unbiased prediction (GBLUP), were used to predict the columnaris disease resistance evaluated by binary survival status. Cross-validation showed that the prediction accuracy of ENGBLUP and GBLUP was 0.7347 and 0.4868, respectively, showing that ENGBLUP had a high prediction accuracy. It was shown that fitting QTL and polygenic effect with different distribution will improve genomic prediction accuracy for binary traits. In this study, an accurate and effective genomic selection method was proposed to predict the columnaris resistance in catfish, and its application should be beneficial to catfish breeding.

BackgroundRecently, there has been a growing interest in the genetic improvement of body measurement traits in farm animals. They are widely used as predictors of performance, longevity, and production traits, and it is worthwhile to investigate the prediction accuracies of genomic selection for these traits. In genomic prediction, the single-step genomic best linear unbiased prediction (ssGBLUP) method allows the inclusion of information from genotyped and non-genotyped relatives in the analysis. Hence, we aimed to compare the prediction accuracy obtained from a pedigree-based BLUP only on genotyped animals (PBLUP-G), a traditional pedigree-based BLUP (PBLUP), a genomic BLUP (GBLUP), and a single-step genomic BLUP (ssGBLUP) method for the following 10 body measurement traits at yearling age of Hanwoo cattle: body height (BH), body length (BL), chest depth (CD), chest girth (CG), chest width (CW), hip height (HH), hip width (HW), rump length (RL), rump width (RW), and thurl width (TW). The data set comprised 13,067 phenotypic records for body measurement traits and 1523 genotyped animals with 34,460 single-nucleotide polymorphisms. The accuracy for each trait and model was estimated only for genotyped animals using five-fold cross-validations.ResultsThe accuracies ranged from 0.02 to 0.19, 0.22 to 0.42, 0.21 to 0.44, and from 0.36 to 0.55 as assessed using the PBLUP-G, PBLUP, GBLUP, and ssGBLUP methods, respectively. The average predictive accuracies across traits were 0.13 for PBLUP-G, 0.34 for PBLUP, 0.33 for GBLUP, and 0.45 for ssGBLUP methods. Our results demonstrated that averaged across all traits, ssGBLUP outperformed PBLUP and GBLUP by 33 and 43%, respectively, in terms of prediction accuracy. Moreover, the least root of mean square error was obtained by ssGBLUP method.ConclusionsOur findings suggest that considering the ssGBLUP model may be a promising way to ensure acceptable accuracy of predictions for body measurement traits, especially for improving the prediction accuracy of selection candidates in ongoing Hanwoo breeding programs.

BackgroundRunning multiple-chain Markov Chain Monte Carlo (MCMC) provides an efficient parallel computing method for complex Bayesian models, although the efficiency of the approach critically depends on the length of the non-parallelizable burn-in period, for which all simulated data are discarded. In practice, this burn-in period is set arbitrarily and often leads to the performance of far more iterations than required. In addition, the accuracy of genomic predictions does not improve after the MCMC reaches equilibrium.ResultsAutomatic tuning of the burn-in length for running multiple-chain MCMC was proposed in the context of genomic predictions using BayesA and BayesCπ models. The performance of parallel computing versus sequential computing and tunable burn-in MCMC versus fixed burn-in MCMC was assessed using simulation data sets as well by applying these methods to genomic predictions of a Chinese Simmental beef cattle population. The results showed that tunable burn-in parallel MCMC had greater speedups than fixed burn-in parallel MCMC, and both had greater speedups relative to sequential (single-chain) MCMC. Nevertheless, genomic estimated breeding values (GEBVs) and genomic prediction accuracies were highly comparable between the various computing approaches. When applied to the genomic predictions of four quantitative traits in a Chinese Simmental population of 1217 beef cattle genotyped by an Illumina Bovine 770 K SNP BeadChip, tunable burn-in multiple-chain BayesCπ (TBM-BayesCπ) outperformed tunable burn-in multiple-chain BayesCπ (TBM-BayesA) and Genomic Best Linear Unbiased Prediction (GBLUP) in terms of the prediction accuracy, although the differences were not necessarily caused by computational factors and could have been intrinsic to the statistical models per se.ConclusionsAutomatically tunable burn-in multiple-chain MCMC provides an accurate and cost-effective tool for high-performance computing of Bayesian genomic prediction models, and this algorithm is generally applicable to high-performance computing of any complex Bayesian statistical model.

Simple SummaryThe objective of our study was to evaluate the predictive ability of a multi-trait genomic prediction model to estimate heritability and genetic correlations of traits such as 305-day milk yield, milk fat percentage, milk protein percentage, milk lactose percentage, and milk dry matter percentage in the Polish Holstein population. Results showed that strong accuracies for the predictions were achieved. The genetic relations with milk yield were negative, as expected, because increasing milk yield decreases the milk components percentage. In conclusion, multi-trait genomic prediction can be more beneficial than single-trait genomic prediction.The objective of our study was to evaluate the predictive ability of a multi-trait genomic prediction model that accounts for interactions between marker effects to estimate heritability and genetic correlations of traits including 305-day milk yield, milk fat percentage, milk protein percentage, milk lactose percentage, and milk dry matter percentage in the Polish Holstein Friesian cow population. For this aim, 14,742 SNP genotype records for 586 Polish Holstein Friesian dairy cows from Poland were used. Single-Trait-ssGBLUP (ST) and Multi-Trait-ssGBLUP (MT) methods were used for estimation. We examined 305-day milk yield (MY, kg), milk fat percentage (MF, %), milk protein percentage (MP, %), milk lactose percentage (ML, %), and milk dry matter percentage (MDM, %). The results showed that the highest marker effect rank correlation was found between milk fat percentage and milk dry matter. The weakest marker effect rank correlation was found between ML and all other traits. Obtained accuracies of this study were between 0.770 and 0.882, and 0.773 and 0.876 for MT and ST, respectively, which were acceptable values. All estimated bias values were positive, which is proof of underestimation. The highest heritability value was obtained for MP (0.3029) and the lowest heritability value was calculated for ML (0.2171). Estimated heritability values were low for milk yield and milk composition as expected. The strongest genetic correlation was estimated between MDM and MF (0.4990) and the weakest genetic correlation was estimated between MY and ML (0.001). The genetic relations with milk yield were negative and can be ignored as they were not significant. In conclusion, multi-trait genomic prediction can be more beneficial than single-trait genomic prediction.

Lack of appropriate donors, non-utilization of high throughput phenotyping and genotyping platforms with high genotype × environment interaction restrained identification of robust QTLs for grain protein content (GPC) in rice. In the present investigation a BC3F4 mapping population was developed using grain protein donor, ARC10075 and high-yielding cultivar Naveen and 190 lines were genotyped using 40 K Affimetrix custom SNP array with the objective to identify stable QTLs for protein content. Three of the identified QTLs, one for GPC (qGPC1.1) and the other two for single grain protein content (qSGPC2.1, qSGPC7.1) were stable over the environments explaining 13%, 14% and 7.8% of the phenotypic variances, respectively. Stability and repeatability of these additive QTLs were supported by the synergistic additive effects of multi-environmental-QTLs. One epistatic-QTL, independent of the main effect QTL was detected over the environment for SGPC. A few functional genes governing seed storage protein were hypothesised inside these identified QTLs. The qGPC1.1 was validated by NIR Spectroscopy-based high throughput phenotyping in BC3F5 population. Higher glutelin content was estimated in high-protein lines with the introgression of qGPC1.1 in telomeric region of short arm of chromosome 1. This was supported by the postulation of probable candidate gene inside this QTL region encoding glutelin family proteins.

With the rapid development of sequencing technology, the application of genomic prediction has become more and more common in breeding schemes of livestocks and crops. Selecting an appropriate statistical model is of central importance to achieve high prediction accuracy. Recently, machine learning models have been expected to upgrade genomic prediction into a new era. However, the perspective still suffers from lack of evidence that machine learning models can generally outperform the traditional ones on empirical data sets. In this study, we compared two machine learning models based on artificial neural network (ANN) and convolutional neural network (CNN) with four traditional models, including genomic best linear unbiased prediction (GBLUP), Bayesian ridge regression (BRR), BayesA and BayesB, using three published data sets for grain yield in wheat. For each model, we considered two variants: modeling and ignoring the genotype-by-environment ([Formula: see text]) interaction. In the comparison, we considered two strategies of cross-validation: predicting genotypes that have not been evaluated in any environment (CV1) and predicting genotypes that have been tested in other environments (CV2). Our results showed that traditional Bayesian models (BayesA, BayesB, and BRR) outperformed GBLUP, ANN and CNN when considering [Formula: see text] interaction. The accuracies of ANN and CNN were higher than traditional models only in CV1 and when [Formula: see text] interaction was ignored. It was also found that the performance of the two machine learning models was significantly affected by the interaction between the CV strategy and the way of treating the [Formula: see text] interaction, while that of the four traditional models was only influenced by whether the [Formula: see text] interaction was considered or not. Thus, machine learning models can be a powerful complementary to the traditional ones and their superiority may depend on the prediction scenario. Among the two machine learning models, we observed that the accuracy of ANN was higher than CNN in most cases, indicating that it is still challenging to adapt complex machine learning models such as CNN to genomic prediction.

Key messageAccurate prediction of genomic breeding values for Timothy was possible using genomic best linear unbiased prediction.Timothy (Phleum pratense L.) is a grass species of great importance for Finnish agricultural production systems. Genotyping-by-sequencing along with genomic prediction methods offer the possibility to develop breeding materials efficiently. In addition, knowledge about the relationships among traits may be used to increase rates of genetic gain. Still, the quality of the genotypes and the validation population may affect the accuracy of predictions. The objectives of the study were (i) to estimate variance components for yield, winter damage and digestibility traits, and (ii) to assess the accuracy of genomic predictions. Variance components were estimated using genomic residual maximum likelihood where the genomic relationship matrix was scaled using a novel approach. Genomic breeding values were estimated using genomic best linear unbiased prediction in single- and multiple-trait settings, and for different marker filtering criteria. Estimates of heritability ranged from 0.13 ± 0.03 to 0.86 ± 0.05 for yield at first cut and organic matter digestibility at second cut, respectively. Genetic correlations ranged from −0.72 ± 0.12 to 0.59 ± 0.04 between yield at first cut and winter damage, and between digestibility at first and second cuts, respectively. Accuracy of prediction was not severely affected by the quality of genotyping. Using family cross-validation and single-trait models, predictive ability ranged from 0.18 to 0.62 for winter damage and digestibility at second cut, respectively. In addition, validation using forward prediction showed that estimated genomic breeding values were moderately accurate with little dispersion. Thus, genomic prediction constitutes a valuable tool for improving Timothy in Finland.

BackgroundPresently, multi-omics data (e.g., genomics, transcriptomics, proteomics, and metabolomics) are available to improve genomic predictors. Omics data not only offers new data layers for genomic prediction but also provides a bridge between organismal phenotypes and genome variation that cannot be readily captured at the genome sequence level. Therefore, using multi-omics data to select feature markers is a feasible strategy to improve the accuracy of genomic prediction. In this study, simultaneously using whole-genome sequencing (WGS) and gene expression level data, four strategies for single-nucleotide polymorphism (SNP) preselection were investigated for genomic predictions in the Drosophila Genetic Reference Panel.ResultsUsing genomic best linear unbiased prediction (GBLUP) with complete WGS data, the prediction accuracies were 0.208 ± 0.020 (0.181 ± 0.022) for the startle response and 0.272 ± 0.017 (0.307 ± 0.015) for starvation resistance in the female (male) lines. Compared with GBLUP using complete WGS data, both GBLUP and the genomic feature BLUP (GFBLUP) did not improve the prediction accuracy using SNPs preselected from complete WGS data based on the results of genome-wide association studies (GWASs) or transcriptome-wide association studies (TWASs). Furthermore, by using SNPs preselected from the WGS data based on the results of the expression quantitative trait locus (eQTL) mapping of all genes, only the startle response had greater accuracy than GBLUP with the complete WGS data. The best accuracy values in the female and male lines were 0.243 ± 0.020 and 0.220 ± 0.022, respectively. Importantly, by using SNPs preselected based on the results of the eQTL mapping of significant genes from TWAS, both GBLUP and GFBLUP resulted in great accuracy and small bias of genomic prediction. Compared with the GBLUP using complete WGS data, the best accuracy values represented increases of 60.66% and 39.09% for the starvation resistance and 27.40% and 35.36% for startle response in the female and male lines, respectively.ConclusionsOverall, multi-omics data can assist genomic feature preselection and improve the performance of genomic prediction. The new knowledge gained from this study will enrich the use of multi-omics in genomic prediction.