Scholarlink

<p>Plant breeding plays a crucial role in addressing the pressing challenges of food insecurity and global hunger, issues that are expected to worsen in the coming years. The development of effective plant breeding pipelines relies on a deep understanding and proficiency in various disciplines, such as phenomics and genomics. Leveraging the five G’s - germplasm characterization, genome assembly, genomic breeding, gene function identification, and gene editing - can significantly boost the pace of crop improvement initiatives. In the past decade, the integration of machine learning (ML) algorithms into the five G’s has gathered increasing attention for their abilities to integrate diverse omics and biological datasets to create precise breeding predictive models. Despite the promise of ML in advancing genomic-assisted breeding, there remains a critical question regarding the extent to which ML can help genomic-assisted breeding and what are their true potentials and efficacies compared to conventional methods. This review evaluates ML’s role in genomics-based plant breeding, highlighting its strengths in prediction accuracy and breeding efficiency but also addressing challenges such as data biases and implementation barriers. It explores applications for improving crop resilience and productivity through the integration of multi-omics data and addressing data biases. Ultimately, the review underscores ML’s potential to transform genomics-based plant breeding while identifying gaps and future research opportunities.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning after a decade: is it still a missing keystone in genomic-based plant breeding?

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年6月4日

查看全文
  • Mohsen Yoosefzadeh-Najafabadi,
  • Alencar Xavier,
  • Milad Eskandari,
  • Mohsen Hesami

摘要

Plant breeding plays a crucial role in addressing the pressing challenges of food insecurity and global hunger, issues that are expected to worsen in the coming years. The development of effective plant breeding pipelines relies on a deep understanding and proficiency in various disciplines, such as phenomics and genomics. Leveraging the five G’s - germplasm characterization, genome assembly, genomic breeding, gene function identification, and gene editing - can significantly boost the pace of crop improvement initiatives. In the past decade, the integration of machine learning (ML) algorithms into the five G’s has gathered increasing attention for their abilities to integrate diverse omics and biological datasets to create precise breeding predictive models. Despite the promise of ML in advancing genomic-assisted breeding, there remains a critical question regarding the extent to which ML can help genomic-assisted breeding and what are their true potentials and efficacies compared to conventional methods. This review evaluates ML’s role in genomics-based plant breeding, highlighting its strengths in prediction accuracy and breeding efficiency but also addressing challenges such as data biases and implementation barriers. It explores applications for improving crop resilience and productivity through the integration of multi-omics data and addressing data biases. Ultimately, the review underscores ML’s potential to transform genomics-based plant breeding while identifying gaps and future research opportunities.