CASE CONTROL BIAS & EFFICIENCY: CANDIDATE GENES & GENE ENVIRONMENT INTERACTION

点击翻译按钮获取中文摘要

In recent years, there has been a focus to understand the etiology of complex diseases which do not follow simple Mendelian, single-locus segregation. Complex diseases are assumed to result from more than one locus and/or environmental factors, or merely exhibit continuous (i.e.quantitative) variation. Quantitative traits have been studied extensively in plant and animal genetics. With the advent of new tools and methods, comprehensive approaches to identify the candidate genes underlying quantitative traits for humans are available. Testing the contribution of candidate genes to quantitative trait variation will become commonplace as more genes are identified. We consider two likelihood-based statistical strategies for testing and quantifying the effect of candidate locus genotypes on a quantitative trait with randomly ascertained pedigree or family data. The first strategy estimates, and then tests the equality of, mean phenotype values association with each genotype. This strategy can be referred to as the "mean effects" strategy. The second strategy estimates and tests a variance component parameter associated with identity-by-state information gathered from alleles at the candidate locus. This strategy tests whether or not allelic variation and co-variation at the locus in question among related individuals explains variation and co-variation in the phenotype of interest among those individuals. This second strategy can be referred as the "variance component" strategy, and forms the basis of most linkage analysis strategies for quantitative traits. Both strategies can be framed in the context of linear models which can accommodate the effects of other factors (e.g. gender, age, sex, etc.) on the phenotype of interest. We consider the use of these models with sib-pair data and bi-allelic loci, and describe analytic derivations that compare and contrast their power and efficiency. Our results suggest that the mean effects model is superior to the variance component model in the sib-pair, di-allelic locus setting.