Bayesian variable selection method for modeling dose-response microarray data under simple order restrictions

Abstract

The aim of the analysis presented below is to investigate dose-response relationship in a microarray setting. Typically, in dose-response experiments the outcome of interest is measured in several (increasing) dose levels and the goal of the analysis is to establish the relationship which represents the dependency of the response on dose. Bayesian modeling of dose-response microarray data offers the possibility to jointly establish the dose response relationships between gene expression and increasing doses of therapeutic compound and to determine the nature of the relationships wherever it exists. The Bayesian variable selection approach provides a modeling framework that allows estimating the posterior probabilities for a given set of pre-specified models and in particular the posterior probability of the model estimated under the null hypothesis of no dose effect. The posterior probabilities are used for multiplicity adjustment using the direct posterior probability approach.