Gene category testing problems involve testing hundreds of null hypotheses that correspond to nodes in a directed acyclic graph. The logical relationships among the nodes in the graph imply that only some configurations of true and false null hypotheses are possible and that a test for a given node should depend on data from neighboring nodes. We use a multivariate nonparametric permutation test to obtain a p-value for each gene category. We then model these p-values with a hidden Markov model that takes the relationships among categories into account. Using a Markov chain Monte Carlo approach, we provide coherent decisions about the differential expression status of each category in this structured multiple hypothesis testing problem. The method - which provides an alternative to get set enrichment analysis and related techniques - will be illustrated by testing Gene Ontology terms for evidence of differential expression. This is joint work with Iowa State PhD Student Kun Liang