STATISTICAL MODELING FOR GENETIC EPIDEMIOLOGY

家族性疾病和人类特征的统计检验

The statistical examination of familial disease, and human traits associated with diseases, considers whether the phenotype under study is familial and, if so, whether some genetic mechanism (e.g. segregation of a major gene and linkage to a known marker) is relevant. Such an analysis should realistically account for less specific sources of familial covariation such as common environment, cultural transmission and other factors with separately indistinguishable effects. Regressive models offer a practical approach to derive the necessary comprehensive models for family data. Details and extensions of the regressive models to be worked out include parsimonious covariate adjustment; multiple logistic models for qualitative traits with more than two categories; association of genotypes of mates; ascertainment corrections; missing data; linkage and other bivariate problems; and variance component and path analytic models. The statistical properties (bias, consistency, diagnostics, likelihood surfaces, and robustness of likelihood ratio tests) will also be studied.