Statistical analysis of microarrays

Microarray statistical analysis involves thousands of hypothesis tests to consider at the same time. Empirical Bayes methods which are well-suited for large scale inference problems seem to be the most appropriate approach for microarray data. In this thesis we describe and compare Efron's ([3],[1],[4]) nonparametric empirical statistical analysis and Newton's and Kendziorski'ร ([12]) parametric empirical statistical analysis on microarray data. Both methods estimate Efron's ([3],[1],[4]) local false discovery rate, which identifies interesting genes and provides information about the power of the experiment.