It is also of current interest to incorporate biological knowledge into microarray data analyses. In many instances, scientists are interested in the effects of a particular gene or pathway on genetic expression. In this context, approaches have been suggested by Zien et al [22] and Pavlidis et al [23] in which biological knowledge as represented by pathway scores or functional annotation status are correlated with gene expression. However, their approaches were univariate. There would be potential gains in efficiencies of analyses by considering joint models for pathways. We are currently studying the applicability of the joint estimation procedure described here to that setting.
