Stan Pounds, Ph.D.

Department of Biostatistics

St. Jude Children’s Research Hospital

Memphis, Tennessee

 

Approximating and partitioning the distribution of p-values arising from microarray studies

 

Microarray technologies allow investigators to simultaneously measure the expression of thousands of genes.  Consequently, the analysis of microarray data consists of thousands of simultaneous hypothesis tests.  Approximating and partitioning the density of p-values for these hypothesis tests is an easily interpreted and implemented method for estimating the occurrence of type I and type II errors in these types of analyses.  This approach is widely applicable because it can easily be used in conjunction with any statistical test that provides a p-value.  The properties of a parametric and a nonparametric approach for estimating the density are explored through simulation.