The Improvement For A Statistical Test And A Sampling Strategy In Linkage Analysis

Linkage analysis is one of the methods used to localize disease traits in the human genome. By estimating and testing the recombination fraction, it searches the evidence of linkage between a known marker system and a putative gene for a disease. Linkage analysis localizes a gene solely on the basis of its location, without regard to its biochemical function.Transmission Disequilibrium Test (TDT) is one of the methods being used. The basic thought of TDT is: if the marker locus is in the neighborhood of the disease locus, then a specific marker allele associated with the disease allele will have higher frequency among diseased individuals compared to normal individuals. With the extent of Transmission Disequilibrium Test (TDT) to a marker locus with multiple alleles, the TDT-like statistical test T , T and T . havebeen suggested. Given a contingency table whose cells are numbers of parents who either do or not transmit the associated marker alleles to affected children, T compares every symmetric pairs, both Tn andT compare row and column total for every marker allele. Tm uses all cells but T uses non-diagonal elements only. Because of some desirable properties of T, it is recommended in the literature. Tmhet, like the other two tests, is a sum of some entries. Every entry has asame weight (coefficient). What we show in this paper (the second part) is that although the sparseness of the contingency table is an aspect that would affect the power of J^, the affect of the weight is also important and should not be ignored. Based on such an idea, a new test, which is a weighted-statistical test of Tmlta, is proposed. We compareits power with that of Tmhet. Almost all simulated results indicate that the new test is better than Tmhet.Another important method of linkage analysis is the Affected Sib Pair (ASP). It is based on the idea that sib pairs with similar phenotypes should have an excess of allele sharing. There are three kinds of sib pairs that can be obtained: both affected member, one and only one affected member, and neither affected member. Blackwelder and Elston (1985) compared three sampling strategies and suggested that the best was sampling from the sib pairs with both affected member. In order to consider all of the sampling strategies, we used the theory of the optimal design for hypothesis test in this paper (the third part) and obtained a different result. Furthermore, we showed that the sampling strategies we obtained was the best.