Random Genetic Mosaics I. Models and Moments
A genetic mosaic is a genetically composite organism, within whose tissues two or more genetically distinct mosaic types of cells coexist. Subsequent to the onset of mosaicism in an individual, the mosaic type of a dividing cell is copied into the two new cells. Mosaic-composition data are measurements of the compositions of one or more tissues in one or more individuals in terms of the mosaic types of their constituent cells. Such data are widely used in studies of tissue development. Their analysis is generally based on assumptions to the effect that mosaic types are randomly assigned to the cells present at the onset of mosaicism, and that the differences between the mosaic types are developmentally unimportant. From precise interpretations of these assumptions, stochastic models of the mosaic composition of a system of tissues will be constructed (for an arbitrary number of mosaic types). The low-order cumulants of the joint distribution of the mosaic-composition variables will be shown to have a simple structure under the stronger models. The data of Nesbitt (1971) will be analyzed. These are measurements of the proportions of type-1 cells in five tissues in 34 mice, with measurement errors distributed as independent binomial proportions. Data-analytic models will be motivated and fitted in which the underlying tissue proportions are given a multivariate normal distribution over mice.
Categorical Data Analysis | Statistical Methodology | Statistical Models | Statistical Theory
Redfearn, William J., "Random Genetic Mosaics I. Models and Moments" (November 1989). U.C. Berkeley Division of Biostatistics Working Paper Series. Working Paper 8.