The Effect of Haplotype Size on Genomic Selection Accuracy and Epistasis: An Empirical Study in Rice

This dataset contains phenotypic and genotypic data collected from rice populations to investigate the impact of haplotype size on genomic selection (GS) accuracy and epistasis. The phenotypic data covers field trials from 2020, 2021, and 2022 for the MP2 and MP6-8 populations. Genotypic data is included for the MP2, MP4, and MP6-8 populations. All scripts necessary for the analysis are also provided, ensuring reproducibility.

Research Hypothesis and Key Findings: Genomic selection (GS) has revolutionized breeding by combining genotype and phenotype data to predict genomic estimated breeding values (GEBVs), potentially accelerating breeding cycles. This study hypothesized that recombination affects haplotype size and linkage disequilibrium (LD), influencing GS prediction accuracy. Specifically, the study aimed to:

Examine the relationship between recombination and haplotype sizes.
Compare additive (A) versus additive + epistasis (A+I) models on prediction accuracy.
Investigate how haplotype resolution in the training set (TS) affects prediction accuracy.
Results showed a direct correlation between LD decay and recombination opportunities within populations, with populations undergoing more recombination displaying smaller haplotype blocks. While the A+I model improved heritability, it did not enhance prediction accuracy. Populations with smaller haplotype sizes in the TS exhibited improved prediction accuracy, highlighting the importance of haplotype size in GS.

Data Description and Use: The phenotypic data includes traits measured across three years of field trials, while the genotypic data represents the underlying genetic makeup of the populations. The unique aspect of this dataset is its focus on populations where recombination rate—and therefore haplotype size—is the primary variable. Researchers can use this dataset to explore the relationship between recombination, haplotype structure, and GS prediction accuracy, providing insights into breeding strategy design.

本数据集收录了从水稻种群中收集的表型及基因型数据，旨在探讨等位基因群大小对基因组选择（GS）准确性和上位性影响的研究。表型数据涵盖了2020年、2021年和2022年MP2及MP6-8种群的田间试验。基因型数据涵盖了MP2、MP4和MP6-8种群。所有分析所需的脚本均已提供，确保了可重复性。

研究假设与主要发现：基因组选择（GS）通过结合基因型和表型数据预测基因组估计育种值（GEBVs），从而革新了育种技术，有望加速育种周期。本研究假设重组影响等位基因群大小和连锁不平衡（LD），进而影响GS预测的准确性。具体而言，研究旨在：

考察重组与等位基因群大小之间的关系。
比较加性（A）与加性+上位性（A+I）模型在预测准确性上的差异。
探究训练集（TS）中等位基因分辨力对预测准确性的影响。
结果表明，连锁不平衡的衰减与种群内重组机会之间存在直接相关性，经历更多重组的种群表现出更小的等位基因群块。尽管A+I模型提高了遗传力，但并未提升预测准确性。在训练集中具有较小等位基因群大小的种群表现出更高的预测准确性，突显了等位基因群大小在GS中的重要性。

数据描述与用途：表型数据包括跨越三年田间试验测量的性状，而基因型数据则代表了种群背后的遗传组成。本数据集的独特之处在于其关注重组率——进而为等位基因群大小——作为主要变量的种群。研究者可以利用本数据集探究重组、等位基因结构与GS预测准确性之间的关系，为育种策略设计提供洞见。