Empirical bayes methods and false discovery rates for microarrays

摘要: In a classic two-sample problem, one might use Wilcoxon's statistic to test for difference between treatment and control subjects. The analogous microarray experiment yields thousands of Wilcoxon statistics, each gene on the array, confronts statistician with difficult simultaneous inference situation. We will discuss two inferential approaches this problem: an empirical Bayes method that requires very little priori Bayesian modeling, frequentist "false discovery rates" proposed by Benjamini Hochberg in 1995. It turns out methods are closely related can be used together produce sensible inferences.