Analysis of genetic-statistical models for Genomic Selection and QTL-Mapping

由于分子遗传学的进步，动物可以一次对数千个标记基因座进行基因分型，提供密集的标记图谱。已经提出使用这些来获得基因组辅助育种值（GABV）。从模拟研究的结果是有前途的高相关性真实和估计的育种值约0.8从单代数据。这与BLUP-EBV对于后代测试的选择候选物的准确性相当。由于一旦基因分型可行，GABV就可用，因此可以缩短育种计划的世代间隔，从而大大增加遗传改良。迄今为止提出的模型假设标记基因座的独立效应，这是不现实的。此外，这些模型捕捉连锁不平衡（LD），而忽略了共分离（CS）的信息。尽管GABV的精确度预计很高，但结合LD和CS的模型应该能够降低基因分型的成本和测试动物的数量。这些模型可能使GABV对猪和家禽养殖更具吸引力。拟议的研究项目的目标是首先开发一个统计模型，其中包括一个依赖结构的影响标记位点。然后，将考虑结合联合收割机LD和CS的模型。这些模型将与模拟数据和家禽养殖公司提供的真实的数据进行评估。

Due to advances in molecular genetics, animals can be genotyped for thousands of maker loci at one time, providing dense marker maps. It has been proposed to use these to obtain genome-assisted breeding values (GABV). Results from simulation studies were promising with high correlations between true and estimated breeding values of around 0.8 from a single generation of data. This is comparable to the accuracy of BLUP-EBV for progeny tested selection candidates. Since GABV are available as soon as genotyping is feasible, generation intervals in breeding programs can be reduced and thus genetic improvement can be increased enormously. The models that have been proposed so far assume independent effects of marker loci, which is not realistic. Furthermore, these models capture linkage disequilibrium (LD) while neglecting information from cosegregation (CS). Even though the precision of GABVs is expected to be high, models combining LD and CS should be able to reduce both the cost for genotyping and the number of test animals. These models might make GABV more attractive for pigs and poultry breeding in particular. The objective of the proposed research project is first to develop a statistical model which includes a dependence structure for effects of marker loci. Then, models that combine LD and CS will be considered. These models will be evaluated with simulated data and real data provided by a poultry breeding company.