A graph modification approach for finding core-periphery structures in protein interaction networks.

Sharon Bruckner,Christian Komusiewicz,Falk Hüffner

doi:10.1186/s13015-015-0043-7

Sharon Bruckner, Christian Komusiewicz + Show 1 more

Open Access

https://doi.org/10.1186/s13015-015-0043-7

Copy DOI

Abstract

The core–periphery model for protein interaction (PPI) networks assumes that protein complexes in these networks consist of a dense core and a possibly sparse periphery that is adjacent to vertices in the core of the complex. In this work, we aim at uncovering a global core–periphery structure for a given PPI network. We propose two exact graph-theoretic formulations for this task, which aim to fit the input network to a hypothetical ground truth network by a minimum number of edge modifications. In one model each cluster has its own periphery, and in the other the periphery is shared. We first analyze both models from a theoretical point of view, showing their NP-hardness. Then, we devise efficient exact and heuristic algorithms for both models and finally perform an evaluation on subnetworks of the S. cerevisiae PPI network.Electronic supplementary materialThe online version of this article (doi:10.1186/s13015-015-0043-7) contains supplementary material, which is available to authorized users.

Highlights

A fundamental task in the analysis of PPI networks is the identification of protein complexes and functional modules
We show an alternative formulation based on the observation that if we correctly guess the partition into core and independent set vertices, we can get a simpler forbidden subgraph characterization for both split cluster graphs and monopolar graphs
The SCAN algorithm [35], like MONOPOLAR EDITING (ME), partitions the graph vertices into “clusters”, which we interpret as cores, and “hubs” and “outliers”, which we interpret as periphery

Summary

Introduction

A fundamental task in the analysis of PPI networks is the identification of protein complexes and functional modules. A basic assumption is that complexes in a PPI network are strongly connected among themselves and weakly connected to other complexes [1]. To obtain a more realistic network model of protein complexes, several approaches incorporate the core–attachment model of protein complexes [2]. In this model, a complex is conjectured to consist of a stable core plus some attachment proteins, which have only transient interactions with the core. The attachment (or: periphery) is less dense, but has edges to one or more cores

Methods

Results

Conclusion