Santantonio Nicholas, Jannink Jean-Luc, Sorrells Mark
Cornell University, Plant Breeding and Genetics Section, School of Integrated Plant Sciences, College of Agriculture and Life Sciences, Ithaca, NY 14853 and
Cornell University, Plant Breeding and Genetics Section, School of Integrated Plant Sciences, College of Agriculture and Life Sciences, Ithaca, NY 14853 and.
G3 (Bethesda). 2019 Mar 7;9(3):675-684. doi: 10.1534/g3.118.200646.
Epistasis is an important contributor to genetic variance. In inbred populations, pairwise epistasis is present as additive by additive interactions. Testing for epistasis presents a multiple testing problem as the pairwise search space for modest numbers of markers is large. Single markers do not necessarily track functional units of interacting chromatin as well as haplotype based methods do. To harness the power of multiple markers while minimizing the number of tests conducted, we present a low resolution test for epistatic interactions across whole chromosome arms. Epistasis covariance matrices were constructed from the additive covariances of individual chromosome arms. These covariances were subsequently used to estimate an epistatic variance parameter while correcting for background additive and epistatic effects. We find significant epistasis for 2% of the interactions tested for four agronomic traits in a winter wheat breeding population. Interactions across homeologous chromosome arms were identified, but were less abundant than other chromosome arm pair interactions. The homeologous chromosome arm pair 4BL/4DL showed a strong negative relationship between additive and interaction effects that may be indicative of functional redundancy. Several chromosome arms appeared to act as hubs in an interaction network, suggesting that they may contain important regulatory factors. The differential patterns of epistasis across different traits demonstrate that detection of epistatic interactions is robust when correcting for background additive and epistatic effects in the population. The low resolution epistasis mapping method presented here identifies important epistatic interactions with a limited number of statistical tests at the cost of low precision.
上位性是遗传变异的一个重要因素。在近交群体中，成对上位性以加性×加性互作的形式存在。由于少量标记的成对搜索空间很大，因此检测上位性会带来多重检验问题。单个标记不一定能像基于单倍型的方法那样追踪相互作用染色质的功能单元。为了在最小化检验次数的同时利用多个标记的功效，我们提出了一种针对整个染色体臂上位性互作的低分辨率检验方法。上位性协方差矩阵由各个染色体臂的加性协方差构建而成。随后，在校正背景加性和上位性效应的同时，利用这些协方差来估计一个上位性方差参数。我们在一个冬小麦育种群体中发现，所检测的四个农艺性状的2%的互作存在显著上位性。鉴定出了同源染色体臂间的互作，但比其他染色体臂对互作的数量要少。同源染色体臂对4BL/4DL在加性效应和互作效应之间呈现出强烈的负相关关系，这可能表明存在功能冗余。几个染色体臂似乎在一个互作网络中充当枢纽，这表明它们可能包含重要的调控因子。不同性状间上位性的差异模式表明，在群体中校正背景加性和上位性效应时，上位性互作的检测是可靠的。这里提出的低分辨率上位性定位方法以低精度为代价，通过有限数量的统计检验识别出重要的上位性互作。
