A Further Study On The Gene Location Of Ordered Trait Locus

In recent years, with the rapid development of statistical genetics, finding the proper statistical methods for mapping multiple quantitative trait loci (QTL) is receiving more and more attention. Among these methods, the multiple-interval mapping (MIM) can map QTL and estimate epistasis simultaneously, but usual-ly it’s aimed at the trait which has a continuous distribution. While the data of ordinal trait bears relatively less information than that of continuous distribution. Therefore, it is more complex to map the QTL of ordinal trait. Because of this particularity, choosing appropriate statistical methods to take into account the dis-tribution of traits is very important for mapping QTL, especially the multiple QTL.In this article, we consider the MIM method in the research on genetic mapping of ordinal traits, assuming that there is an underlying response variable called lia-bility, which is connected with ordinal trait values through a threshold model, then connecting the liability and QTL genotypes through a genetic and statistical models. Finally we can set up a cumulative logistic regression model to express the relation-ship between the ordinal data and QTL genotypes. However, when considering this kind of problems of mapping QTL, most people use the method of genome-wide scan. In order to make the interval mapping more straightforward, we treat the recombination rates as unknown parameters, and then obtain the estimates of QTL position parameters, threshold model parameters and the QTL effect parameters simultaneously via the EM algorithm. In this article, we do some simulation exper-iments according to the different situations of interference coefficient. What’s more, we also choose an example to make hypothesis tests of the reasonalbeness of model and estimate both the parameters of the model and the positions of QTLs. Both the results of simulations and example show that the approaches we proposed are reasonable and effective.