Mining Multiple Biological Data for Reconstructing Signal Transduction Networks

Abstract

Signaling transduction networks (STNs) are the key means by which a cell converts an external signal (e.g. stimulus) into an appropriate cellular response (e.g. cellular rhythms of animals and plants). The essence of STN is underlain in some signaling features scattered in various data sources and biological components overlapping among STN. The integration of those signaling features presents a challenge. Most of previous works based on PPIs for STN did not take the signaling properties of signaling molecules and components overlapping among STN into account. This paper describes an effective computational method that can exploit three biological facts of STN applied to human: protein-protein interaction networks, signaling features and sharing components. To this end, we introduce a soft-clustering method for doing the task by exploiting integrated multiple data, especially signaling features, i.e., protein-protein interactions, signaling domains, domain-domain interactions, and protein functions. The gained results demonstrated that the method was promising to discover new STN and solve other related problems in computational and systems biology from large-scale protein interaction networks. Other interesting results of the early work on yeast STN are additionally presented to show the advantages of using signaling domain-domain interactions.