TINTIN: Exploiting Target Features for Signaling Network Similarity Computation and Ranking

Network similarity ranking attempts to rank a given set of networks based on its "similarity" to a reference network. State-of-the-art approaches tend to be general in the sense that they can be applied to networks in a variety of domains. Consequently, they are not designed to exploit domain-specific knowledge to find similar networks although such knowledge may yield interesting insights that are unique to specific problems, paving the way to solutions that are more effective. We propose Tintin which uses a novel target feature-based network similarity distance for ranking similar signaling networks. In contrast to state-of-the-art network similarity techniques, Tintin considers both topological and dynamic features in order to compute network similarity. Our empirical study on signaling networks from BioModels with real-world curated outcomes reveals that Tintin ranking is different from state-of-the-art approaches.