A review of deep learning applications for genomic selection

BMC Genomics, Journal Year: 2021, Volume and Issue: 22(1)

Published: Jan. 6, 2021

Several conventional genomic Bayesian (or no Bayesian) prediction methods have been proposed including the standard additive genetic effect model for which variance components are estimated with mixed equations. In recent years, deep learning (DL) considered in context of prediction. The DL nonparametric models providing flexibility to adapt complicated associations between data and output ability very complex patterns.

Language: Английский

---

Machine learning approaches for crop improvement: Leveraging phenotypic and genotypic big data

Journal of Plant Physiology, Journal Year: 2020, Volume and Issue: 257, P. 153354 - 153354

Published: Dec. 29, 2020

Highly efficient and accurate selection of elite genotypes can lead to dramatic shortening the breeding cycle in major crops relevant for sustaining present demands food, feed, fuel. In contrast classical approaches that emphasize need resource-intensive phenotyping at all stages artificial selection, genomic dramatically reduces phenotyping. Genomic relies on advances machine learning availability genotyping data predict agronomically phenotypic traits. Here we provide a systematic review applied single multiple traits past decade. We gather intermediate phenotypes, e.g. metabolite, protein, gene expression levels, along with developments modeling techniques further improvements selection. addition, critical view factors affect attention transferability models between different environments. Finally, highlight future aspects integrating high-throughput molecular from omics technologies biological networks crop improvement.

Language: Английский

---

Multitrait machine‐ and deep‐learning models for genomic selection using spectral information in a wheat breeding program

The Plant Genome, Journal Year: 2021, Volume and Issue: 14(3)

Published: Sept. 5, 2021

Abstract Prediction of breeding values is central to plant and has been revolutionized by the adoption genomic selection (GS). Use machine‐ deep‐learning algorithms applied complex traits in plants can improve prediction accuracies. Because tremendous increase collected data programs slow rate genetic gain increase, it required explore potential artificial intelligence analyzing data. The main objectives this study include optimization multitrait (MT) models for predicting grain yield protein content wheat ( Triticum aestivum L.) using spectral information. This compares performance four deep‐learning‐based unitrait (UT) MT with traditional best linear unbiased predictor (GBLUP) Bayesian models. dataset consisted 650 recombinant inbred lines (RILs) from a spring program grown three years (2014–2016), were at heading filling stages. MT‐GS performed 0–28.5 −0.04 15% superior UT‐GS Random forest multilayer perceptron performing predict both traits. Four explored gave similar accuracies, which less than increased computational time. Green normalized difference vegetation index (GNDVI) predicted seven out nine Overall, concluded that accuracy should be employed large‐scale programs.

Language: Английский

---

Need for speed: manipulating plant growth to accelerate breeding cycles

Current Opinion in Plant Biology, Journal Year: 2021, Volume and Issue: 60, P. 101986 - 101986

Published: Jan. 9, 2021

Language: Английский

---

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

Journal of Applied Genetics, Journal Year: 2024, Volume and Issue: 65(2), P. 225 - 240

Published: Jan. 13, 2024

Language: Английский

---

Enhancing genomic‐based forward prediction accuracy in wheat by integrating UAV‐derived hyperspectral and environmental data with machine learning under heat‐stressed environments

The Plant Genome, Journal Year: 2025, Volume and Issue: 18(1)

Published: Jan. 8, 2025

Abstract Integrating genomic, hyperspectral imaging (HSI), and environmental data enhances wheat yield predictions, with HSI providing detailed spectral insights for predicting complex grain (GY) traits. Incorporating single nucleotide polymorphic markers (SNPs) resulted in a substantial improvement predictive ability compared to the conventional genomic prediction models. Over course of several years, varied due diverse weather conditions. The most comprehensive parametric model tested, which included SNPs, HSI, covariates data, consistently achieved best results, closely followed by machine learning (ML) approaches when considering same omics data. For example, (M9), under forward cross‐validation scheme, predicted GY 2023 growing season using from 2021 2022 correlation between observed values 0.53. This demonstrated superior performance less models, emphasizing advantage integrating numerous sources their interactive effects. Furthermore, comparing top 25% lines versus corresponding highest GY, M9 returned coincide index (CI) 55% (i.e., both sets, were common), whereas performing ML (gradient boosting regression), CI was 46%. study highlights potential multi‐data source accelerate selection heat‐tolerant genotypes.

Language: Английский

---

Harness the power of genomic selection and the potential of germplasm in crop breeding for global food security in the era with rapid climate change

The Crop Journal, Journal Year: 2020, Volume and Issue: 8(5), P. 688 - 700

Published: June 5, 2020

Crop genetic improvements catalysed population growth, which in turn has increased the pressure for food security. We need to produce 70% more meet demands of 9.5 billion people by 2050. Climate changes have posed challenges global supply, while narrow base elite crop cultivars further limited our capacity increase gain through conventional breeding. The effective utilization resources germplasm collections improvement is crucial increasing address supply. Genomic selection (GS) uses genome-wide markers and phenotype information from observed populations establish associations, followed predict phenotypic values test populations. Characterizing an extensive collection can serve a dual purpose GS, as reference predicting model, mining desirable variants incorporation into cultivars. New technologies, such high-throughput genotyping phenotyping, machine learning, gene editing, great potential contribute genome-assisted Breeding programmes integrating characterization, GS emerging technologies offer promise accelerating development with improved yield enhanced resistance tolerance biotic abiotic stresses. Finally, scientifically informed regulations on new breeding sharing resources, genomic data, bioinformatics expertise between developed developing economies will be key meeting rapidly changing climate demand food.

Language: Английский

---

Multi-Trait Multi-Environment Genomic Prediction of Agronomic Traits in Advanced Breeding Lines of Winter Wheat

Frontiers in Plant Science, Journal Year: 2021, Volume and Issue: 12

Published: Aug. 18, 2021

Genomic prediction is a promising approach for accelerating the genetic gain of complex traits in wheat breeding. However, increasing accuracy (PA) genomic (GP) models remains challenge successful implementation this approach. Multivariate have shown promise when evaluated using diverse panels unrelated accessions; however, limited information available on their performance advanced breeding trials. Here, we used multivariate GP to predict multiple agronomic 314 and elite lines winter 10 site-year environments. We multi-trait (MT) model with two cross-validation schemes representing different scenarios (CV1, completely unphenotyped lines; CV2, partially phenotyped correlated traits). Moreover, extensive data from multi-environment trials (METs) were cross-validate Bayesian (MTME) that integrates analysis multiple-traits, such as G × E interaction. The MT-CV2 outperformed all other predicting grain yield significant improvement PA over single-trait (ST-CV1) model. MTME performed better traits, average ST-CV1 reaching up 19, 71, 17, 48, 51% yield, protein content, test weight, plant height, days heading, respectively. Overall, empirical analyses elucidate potential both are training population related preliminary lines. Further, practical application program reduce phenotyping cost sparse testing design. This showed complementing METs can substantially enhance resource efficiency. Our results demonstrate GS great implementing programs.

Language: Английский

---

Multi-Trait Multi-Environment Genomic Prediction for End-Use Quality Traits in Winter Wheat

Frontiers in Genetics, Journal Year: 2022, Volume and Issue: 13

Published: Jan. 31, 2022

Soft white wheat is a class used in foreign and domestic markets to make various end products requiring specific quality attributes. Due associated cost, time, amount of seed needed, phenotyping for the end-use trait delayed until later generations. Previously, we explored potential using genomic selection (GS) selecting superior genotypes earlier breeding program. Breeders typically measure multiple traits across locations, it opens up avenue exploring multi-trait–based GS models. This study’s main objective was explore multi-trait models predicting seven different cross-validation, independent prediction, across-location predictions The population consisted 666 soft planted 5 years at two locations Washington, United States. We optimized compared performances four uni-trait– models, namely, Bayes B, best linear unbiased prediction (GBLUP), multilayer perceptron (MLP), random forests. accuracies were 5.5 7.9% uni-trait within-environment predictions. Multi-trait machine deep learning performed GBLUP B predictions, but their advantages diminished when genotype by environment component included model. highest improvement accuracy, that is, 35% obtained flour protein content with MLP study showed enhance accuracy information from previously phenotyped traits. It would assist speeding cycle time cost-friendly manner.

Language: Английский

---

Multi-trait ensemble genomic prediction and simulations of recurrent selection highlight importance of complex trait genetic architecture for long-term genetic gains in wheat

in silico Plants, Journal Year: 2023, Volume and Issue: 5(1)

Published: Jan. 1, 2023

Abstract Cereal crop breeders have achieved considerable genetic gain in genetically complex traits, such as grain yield, while maintaining diversity. However, focus on selection for yield has negatively impacted other important traits. To better understand multi-trait within a breeding context, and how it might be optimized, we analysed genotypic phenotypic data from diverse, 16-founder wheat multi-parent advanced generation inter-cross population. Compared to single-trait models, ensemble genomic prediction models increased accuracy almost 90 % of improving by 3–52 %. For non-parametric (Random Forest) also outperformed simplified, additive (LASSO), increasing 10–36 Simulations recurrent then showed that sustained greater forward optimized long-term gains. found indirect responses related involving antagonistic trait relationships. We indices could effectively optimize undesirable relationships, the trade-off between protein content, or combine traits interest, weed competitive ability. including Random Forest rather than LASSO true model accelerated extended whilst These results (i) suggest roles pleiotropy epistasis wider context programmes, (ii) provide insights into mechanisms continued limited genepool optimization multiple improvement.

Language: Английский

---