Parameter estimation for a finite mixture model in high dimensional applications

Finite mixture models have found use in the analysis of high dimensional data such as result from microarray experiments. A common goal of a microarray experiment is to identify genes that express differentially between two types of tissues or between two experimental conditions. Some investigators found that the distribution of P-values from tests for differential genetic expression contains useful information regarding several quantities of interest. A uniform-beta mixture distribution (mix-o-matic) has been employed to model this distribution...This dissertation covers three topics: 1) the performance of interval estimates of model parameters using three computational methods including a comparison of the computational methods; 2: a relatively recent approach based on a number theoretic method for obtaining MLEs, its extensions and a comparison to Newton-type methods; 3) FDR estimation in the mix-o-matic and a comparison with eight other techniques for estimating FDR, all techniques making use of information in the distribution of P-values --Abstract, page iii