Abstract:
|
Over the last decade, large-scale multiple testing has found itself at the forefront of modern data analysis. In many applications data are correlated, so that the observed test statistic used for detecting a non-null case, or signal, at each location in a dataset carries some information about the chances of a true signal at other locations. Brown et al. (2014) proposed a Bayesian multiple testing model that accounts for dependence through a CAR model. Here, we propose more general definitions of neighborhood structures that allow for inclusion of points with no neighbors at all, something that is not possible under conventional CAR models. This modification allows for the simultaneous modeling of dependent and independent cases, resulting in increased precision in the estimates of non-null signal strengths. We illustrate the effectiveness and applicability of our proposed model by using it to analyze both simulated and real microarray data in which the genes exhibit nontrivial dependence that is determined by physical adjacency on a chromosome or predefined gene pathways.
|