|
Abstract:
|
The identification of rare genetic variants has revived the use and analysis of related individuals in genetic and genomic studies. However, unmeasured environmental factors segregate from generation to generation in much the same way as genes consequently, genetic studies are often marred by inadequate statistical power and complications due to gene-environment interactions. Current statistical models typically use random intercepts in generalized linear mixed models (GLMM) to account for unmeasured environmental effects, but random intercepts are seldom realistic in larger families and the performance and power of GLMMs may be greatly improved by enhanced modeling of the unmeasured environmental effects. We present a flexible three-parameter model for the environmental relationship among relatives and show - through analytical results and simulations - the nice performance of the proposed method under various genetic models. We also show that the classical random intercept is a special, nested cased of the model, and that it becomes possible to separate environmental factors between spouses and between parent-offspring.
|
