Finding Conserved Patterns in Multilayer Networks.

Motivation: Traditional methods often represent complex systems as a single, static, and binary network. These models are inadequate in capturing complex cellular interactions which vary under different conditions as well as over time. Furthermore the same set of molecules can interact in varying patterns across different interactomes. In this paper, we model cellular interactions as a set of network topologies, called multilayer networks. We consider motif counting, one of the most fundamental problems in network analysis. Existing motif counting and identification methods are limited to single network topologies, and thus they cannot be directly applied on multilayer networks. Results: In this paper, we extend the classical network motif identification problem to multilayer networks. We develop an efficient and accurate method to solve this problem. Our results on Escherichia coli (E.coli) transcription regulatory network under different experimental conditions show that our method scales to real networks and more importantly can uncover conserved functional characteristics of genes participating in the network under various conditions with very low false discovery rates.