Research On Statistical Methods For Gene Mapping

Gene mapping is an important study topic in bioinformatics and statisticalgenetics. So far, researchers have found many disease loci, most of which are singlegene disease loci, in other words, the study about complex disease is still a di?culty.In statistical genetics, the methods for gene mapping mainly includes twoclasses, i.e., linkage analysis method and association analysis method. The formermethod can infer the relative location between the gene locus of interest and someknown marker loci using the genotype data and phenotype data. The relative loca-tion can be described by the parameter of recombination fraction, because recom-bination fraction and the genetic distance can be transformed mutually. Therefore,one of the important parameter we need infer is the recombination fraction. Ofcourse, this belongs to the parameter method. The non-parameter method willnot need inferring parameters. In recent years, with the completeness of humangenome project, the methods used in constructing genetic map and gene mappingmake many progress.Association analysis is widely used in epidemiology, which is an e?ective wayto find the locations of complex disease genes. The association between geneticmarkers and disease means that the genetic markers may exist strong linkage dis-equilibrium with disease genes, or the genetic markers may exist association withoccurrence of disease. The association degree is usually described by the linkagedisequilibrium, which is a population property. The linkage disequilibrium degreecan be inferred by population data. The study about this aspect is called asso-ciation analysis in literatures. In association analysis, by the association state ofadjacent loci, we can detect the disease loci. Once the accurate location or ancertain region is obtained, the curing of disease will become easy.In the dissertation, we first propose a new association method (RTTFP), whichis an improvement of an existed method. The new method detect the disease loci with a certain power, and outperform current methods (TTFP and TTFPBFA).At the same time, the new method can control the population stratification.Second, to a known dataset, we propose a method to detect gene-gene, andgene-environment interaction, in which, a key stage is collapsing the rare variantsin a gene to a combined are variant (CRV) if the rare variant is included. Thenew variant is used to analyze. We use the new method to the GAW17 dataset toidentify the interaction between KDR gene and smoking status for the quantitativetrait Q1.Finally, we make some study in the gene mapping of quantitative trait loci(QTL). We consider the problem of QTL mapping when the number of markersis larger than the sample size, and propose a new idea for dealing with the prob-lem. We verify the feasibility of the new method by example analysis and simula-tion study, and discuss the advantageousness and disadvantageousness of the newmethod with the current ones. Theoretical analysis and simulation study showthat the new method is feasible in practice, and can be used in the genome-wisedetecting of QTL.