Network models are an invaluable tool for integrating multiple data types and for modeling interactions between biological elements. One common question that arises, however, is what to do with such a network beyond making a pretty picture. In this talk, I will describe two applications in which we have used network structure to explain features of biological systems. In the first, we construct bipartite eQTL networks from genotype and gene expression data collected by the GTEx consortium. In the second, we explore the response of gene regulatory networks in Mycobacterium tuberculosis to a targeted drug treatment. In both cases, we find that the network structural properties reflect constraints operating on each biological system, and that each network has its own unique, informative feature set. For example, in the eQTL networks, we find the global hubs are not very informative but degree is very useful in interpreting the tuberculosis regulatory network. I will also discuss statistical testing and validation of these networks using functional annotation and wet lab experiments.