PMC:1481596 / 28074-30130 JSONTXT

{"target":"https://pubannotation.org/docs/sourcedb/PMC/sourceid/1481596","sourcedb":"PMC","sourceid":"1481596","source_url":"https://www.ncbi.nlm.nih.gov/pmc/1481596","text":"In this work we proposed a tree representation for protein interaction graphs called Tree of Complexes representation. Nodes in the Tree of Complexes are functional groups and the tree satisfies the additional condition that functional groups that contain any fixed protein form a connected subgraph. In this way, our representation captures not only the overlap between functional groups but, potentially, also the manner in which proteins enter and leave their enclosing functional groups. We developed a method (together with the corresponding graph-theoretical theory) for efficient identification of such overlapping functional groups and construction of the corresponding Tree of Complexes. In particular, our method differs from other approaches in that it does not attempt to enumerate disjoint complexes but instead identifies and represents relations between overlapping functional groups. Even though the Tree of Complexes representation is not unique, the protein interaction networks that we analyzed admit very few alternative tree topologies. If we ask for a tree topology with a maximum number of leaves, as not to impose an artificial order between functional groups, the number of tree topologies is reduced even further. Thus, in the TNFα/NF-κB signalling pathway this results in a unique Tree of Complexes representation and in the pheromone signalling pathway in two very closely related possible Tree of Complexes representations. The nature of high-throughput protein interaction data does not directly imply that this data encodes temporal relations. We demonstrated that our method is frequently capable of discovering such temporal relations. Interestingly, temporal associations can also be implicated in the absence of actual interaction in the data. For example, in the case of the pheromone signaling pathway, our method correctly included KSS1 and FUS3 in the same functional group (treated here as temporal associations), despite the fact that there is no link between KSS1 and FUS3 in the input protein interaction network.","tracks":[]}