From DNA to harvest: genomic prediction sharpens rapeseed breeding accuracy

Accurately predicting complex agronomic traits remains a major bottleneck in crop breeding. This study demonstrates how optimized genomic prediction models can reliably forecast flowering time, yield components, and oil content in rapeseed using genome-wide genetic information. By integrating trait-associated genetic variants with both traditional statistical models and machine-learning approaches, the research achieves high prediction accuracy for multiple economically important traits. The results show that incorporating genome-wide association signals substantially enhances model performance, particularly for flowering time and thousand-seed weight. These findings highlight the potential of genomic prediction to accelerate selection decisions, reduce breeding cycles, and enable simultaneous improvement of multiple traits in oilseed crops.

Rapeseed is one of the world’s most important oil crops, yet its breeding progress is constrained by the genetic complexity of traits such as flowering time, seed yield, and oil content. These traits are controlled by many genes with small individual effects and are strongly influenced by population structure and domestication history. Traditional breeding and marker-assisted selection often struggle to capture this complexity efficiently. Meanwhile, whole-genome sequencing now provides abundant genetic data, but using this information effectively for trait prediction remains challenging. Based on these challenges, there is a need to develop and optimize robust genomic prediction strategies to support rapid, accurate selection of complex traits in rapeseed breeding.

A research team from the Oil Crops Research Institute of the Chinese Academy of Agricultural Sciences and collaborating institutions reported (DOI: 10.1093/hr/uhaf115) on April 30, 2025, in Horticulture Research, a comprehensive genomic prediction framework for rapeseed improvement. Using a globally diverse panel of 404 rapeseed breeding lines, the study systematically evaluated traditional genomic models and machine-learning approaches to predict flowering time, yield-related traits, and oil content. The work provides a practical roadmap for integrating genomic prediction into modern rapeseed breeding programs.

The researchers analyzed high-density genome resequencing data from 404 rapeseed accessions representing spring, winter, and semiwinter ecotypes worldwide. After identifying more than 23 million high-quality genetic variants, they combined phenotypic data collected over two growing seasons with genome-wide association studies to detect key loci controlling flowering time, seed yield components, and oil content. Twenty-two significant quantitative trait loci were identified, many of which overlapped across traits, reflecting shared genetic regulation.

To evaluate prediction performance, seven genomic prediction models—including GBLUP, Bayes–Lasso, and several machine-learning algorithms—were tested using different sets of genetic features. Models incorporating GWAS-associated variants consistently outperformed those using randomly selected or dimension-reduced markers. Prediction accuracy exceeded 90% for flowering time and thousand-seed weight, while yield and oil-related traits reached accuracies above 80%. Traditional models such as GBLUP and Bayes–Lasso showed strong and stable performance, particularly when sample size was moderate, whereas machine-learning models performed competitively under certain feature-selection strategies. Importantly, the study showed that including both major and minor-effect variants maximized predictive power while reducing genotyping costs.

According to the researchers, genomic prediction offers a powerful solution to the long-standing challenge of improving complex traits controlled by many genes. By carefully selecting trait-associated genetic variants and matching them with appropriate prediction models, breeders can reliably forecast performance before plants reach maturity. This approach not only shortens breeding cycles but also enables simultaneous optimization of flowering time, yield, and oil quality, which are often difficult to improve together using conventional methods.

The optimized genomic prediction framework described in this study can be readily integrated into rapeseed breeding programs worldwide. By reducing reliance on lengthy field evaluations, breeders can make earlier and more precise selection decisions, accelerating genetic gain. The approach also supports cost-effective genotyping strategies, making large-scale genomic selection more accessible. Beyond rapeseed, the study provides a transferable methodology for other crops with complex trait architectures, offering a scalable pathway toward data-driven, genome-enabled breeding that meets growing global demands for edible oils and agricultural sustainability.

###

References

DOI

10.1093/hr/uhaf115

Original Source URL

https://doi.org/10.1093/hr/uhaf115

Funding information

This research was supported by the National Key Research and Development Program of China (2023YFD120140203), the National Natural Science Foundation of China (U19A2029), the earmarked fund for CARS (CARS-12); the Science and Technology Innovation Project of the Chinese Academy of Agricultural Sciences and Shannan City Municipal Science and Technology Program Projects (SNSBJKJJHXM2023001) are also acknowledged.

About Horticulture Research

Horticulture Research is an open access journal of Nanjing Agricultural University and ranked number one in the Horticulture category of the Journal Citation Reports ™ from Clarivate, 2023. The journal is committed to publishing original research articles, reviews, perspectives, comments, correspondence articles and letters to the editor related to all major horticultural plants and disciplines, including biotechnology, breeding, cellular and molecular biology, evolution, genetics, inter-species interactions, physiology, and the origination and domestication of crops.