Topology Search over Biological Databases

We introduce the notion of a data topology and the problem of topology search over databases. A data topology summarizes the set of all possible relationships that connect a given set of entities. Topology search enables users to search for data topologies that relate entities in a large database, and to effectively summarize and rank these relationships. Using topology search over a biological database, users can ask, for example, how transcription factor proteins are related to DNAs in humans. However, detecting topologies in large databases is a difficult problem because entities can be connected in multiple ways. In this paper, we formalize the notion of data topologies, develop efficient algorithms for computing data topologies based on user queries, and evaluate our algorithms using a real biological database, the Biozon database (www.biozon.org).