Efficiency Of Selective Genotyping In Quantatitive Trait Loci (qtl) Mapping

Most important agronomic and quality traits in crop breeding are quantitative traits, which are controlled by multiple genes, known as quantitative trait loci (or QTL), and can be easily affected by environments. QTL mapping is an important step in gene fine mapping, map-based cloning, and the efficient use of gene information in molecular breeding. Rapid increase in the availability of fine-scale genetic marker maps has led to the intensive use of QTL mapping in the genetic study of quantitative traits. Through QTL mapping, quantitative traits genes, like qualitative traits, can also be located on chromosomes with individual genetic effects being estimated. Results from QTL mapping can largely assist breeders acquire the gene information on important breeding traits, select and trace important genes though closely linked molecular markers, which in turn improving the prediction precision and the selection efficiency.Most QTL mapping and fine mapping studies are based on genotyping an entire germplasm collection or segregating population with marker evenly covering the whole genome. The entire-population approach is extensive, time-consuming and expensive when mapping minor genes where a large size of population is needed. In comparison, selective genotyping requires genotyping only part of individuals from the high and/or low tails of the phenotypic distribution across the germplasm collection or segregating population. Several studies have shown that selective genotyping is an efficient method to locate major QTL. However, there are few studies on the relative efficiency of selective genotyping compared with the entire population approaches. Major objective of this study was to use computer simulation to better understand the effects of potential confounding factors on the QTL detection power of selective genotyping, including entire population size, relative and absolute tail size, and marker density under various genetic models by comparison with other QTL mapping methods based on entire population analysis (inclusive composite interval mapping, ICIM and simple interval mapping, SIM). Genetic models considered in this study included linkage and interaction effects between target QTL as well as different levels of the phenotypic variation explained by the target QTL. Several real populations were also used to compare selective genotyping with other entire-population mapping methods. Estimation of QTL effects in selective genotyping was briefly investigated at the end of this study. Major results from this study were summarized as follows.1. Effect of marker density, population size and selected proportion on QTL detection power of selective genotyping: Detection power of independent QTL was not directly affected by marker density but by the distance between QTL and its nearest marker. QTL will be detected with high power when one QTL was closely linked some markers, though only sparse markers covering the whole genome. The increase in whole population size could improve the QTL detection power. Sizes 15-35% in each tail resulted in the highest power.2. Effect of linkage distance on QTL detection power of selective genotyping: Linked QTL could be detected by increasing marker density and population size in selective genotyping. Each of the linked QTL could be detected when there was at least one empty marker interval. However, if two QTL were tightly linked, such as linkage distance was below 5 cM, selective genotyping was less efficient. But this may be case for other mapping methods. When the linkage distance was greater than 60 cM, selective genotyping resulted in similar powers for linked and independent QTL.3. Detection of epistatic QTL in selective genotyping: Selective genotyping could detect epistatic QTL when both QTL had significant additive effects. Selective genotyping was unable to detect two interacting QTL where the individuals additive effects were not present, which may represent a disadvantage of selective genotyping in QTL mapping.4. Comparison of selective genotyping with entire-population approaches in actual populations: In RIL, DH and BC populations, similar results were observed as have been seen in simulated populations. Selective genotyping detected both independent and linked QTL. Selective genotyping also detected sparsely linked QTL when the entire population size was 100-300. In F2 population, the genetic model was more complex due to present of dominance effect. Selective genotyping only detected part of QTL with significant additive effects. QTL with partial to complete to over dominance effects were not properly detected, indicating that selective genotyping may be less efficient with F2 populations, compared with entire-population approaches. Additional simulations with F2 populations may be needed to validate this observation. In all actual populations, QTL detected by two-tail selective genotyping is more consistent with entire-population approaches than those by one-tail selective genotyping.5. Estimation of QTL effect in selective genotyping: Relatively small QTL effect can be precisely estimated in selective genotyping by using mean difference between the two genotypes in the two pooled-tails or selection coefficient. Large QTL effect is overestimated by this method. Therefore, a better method of estimating QTL effect in selective genotyping may still be needed.