Department of Agronomy and Horticulture, University of Nebraska-Lincoln, Lincoln, NE, USA.
Center for Plant Science Innovation, University of Nebraska-Lincoln, Lincoln, NE, USA.
Plant Biotechnol J. 2019 May;17(5):893-905. doi: 10.1111/pbi.13023. Epub 2018 Nov 9.
Association studies use statistical links between genetic markers and the phenotype variation across many individuals to identify genes controlling variation in the target phenotype. However, this approach, particularly conducted on a genome-wide scale (GWAS), has limited power to identify the genes responsible for variation in traits controlled by complex genetic architectures. In this study, we employ real-world genotype datasets from four crop species with distinct minor allele frequency distributions, population structures and linkage disequilibrium patterns. We demonstrate that different GWAS statistical approaches provide favourable trade-offs between power and accuracy for traits controlled by different types of genetic architectures. FarmCPU provides the most favourable outcomes for moderately complex traits while a Bayesian approach adopted from genomic prediction provides the most favourable outcomes for extremely complex traits. We assert that by estimating the complexity of genetic architectures for target traits and selecting an appropriate statistical approach for the degree of complexity detected, researchers can substantially improve the ability to dissect the genetic factors controlling complex traits such as flowering time, plant height and yield component.
关联研究利用遗传标记与许多个体表型变异之间的统计联系，来识别控制目标表型变异的基因。然而，这种方法，特别是在全基因组范围内（GWAS）进行，对于识别控制复杂遗传结构控制的性状的基因的能力有限。在这项研究中，我们利用来自四个具有不同次要等位基因频率分布、群体结构和连锁不平衡模式的作物物种的真实基因型数据集。我们证明，不同的 GWAS 统计方法在由不同类型的遗传结构控制的性状的功效和准确性之间提供了有利的权衡。FarmCPU 为中等复杂的性状提供了最有利的结果，而从基因组预测中采用的贝叶斯方法为极复杂的性状提供了最有利的结果。我们断言，通过估计目标性状遗传结构的复杂性，并根据检测到的复杂程度选择适当的统计方法，研究人员可以大大提高解析控制开花时间、植物高度和产量组成等复杂性状的遗传因素的能力。
