Simulation Study On Experiment Design For Association Mapping

This thesis targets on exploring the performance of common experiment designs for association mapping with Monte-Carlo simulation technology, aiming to clearly understand the applicability of half-diallel cross designs and its modified types, nested-association mapping population design for association mapping. Also it simulated the influence of the experiment design related factors, such as sample size, mating types, the number of founder parents, the crossing generation as well as rare variants. The main conclusions are as follows:(1) Commonly used half-diallel cross design, nested-association mapping design, and their modified types can be employed for association study. Amongst, half-diallel cross design and factorial design are the most cost-effective and efficient experiment designs, followed by the independent round robin design and the nested association mapping design. On the contrary, the double round robin design and the single round robin design are not suggested for association mapping because of their low mapping power and biased estimates in effects.(2) Under the heritability of 40% and at least 2,250 sample size, with the increasing of founder parents number, the mapping power of the half-diallel cross design showed an increasing tendency and almost reached 100% for most QTS; while, and the unbiasedness of effect estimate was approximately achieved. The sample size of 2250 can also be regarded as reference for other design. The individual line selection in experiment should be carried out as early as possible to capture the most polymorphisms, and the early generation populations, such as the population consisting of F2 of each bi-parental combination, are preferred for obtaining better results of association mapping. Simulation showed that the natural population could be used to analyze rare variants with relatively larger genetic effects, however, the estimation of effect is unstable.Some results of this research could be used to guide designing a cost-effective experiment for association mapping to promote the efficiency of the dissection of quantitative genetic architecture and functional gene mining.