This thesis presents novel computational approaches to infer biomolecular systems for synaptic specificity. The specific issues addressed in this thesis are the identification of biologic mechanisms that select mutually exclusive alternatively spliced exons, and, the integration of neuronal connectivity information with gene expression data to identify synergistic gene sets associated with synapses. We use the well-described neural network of the model C. elegans organism to outline an information-theoretic methodology that not only identifies gene sets most associated with the phenotype ab initio, but also provides insight into the way in which these genes may interact with each other in the context of the specific phenotype. This systems-based methodology combines the modeling power of multivariate analysis along with the descriptive power of decomposing gene sets into interacting components that can help provide insights into molecular pathways associated with biological outcomes.