Finding Collections of Protein Modules in Protein-Protein Interaction Networks

Pierre-Nicolas Mougel (1, 2) , Christophe Rigotti (1, 2, 3) , Olivier Gandrillon (4, 5)

1 LIRIS - Laboratoire d'InfoRmatique en Image et Systèmes d'information
2 BEAGLE - Artificial Evolution and Computational Biology
3 DM2L - Data Mining and Machine Learning
4 DRACULA - Multi-scale modelling of cell dynamics : application to hematopoiesis
5 CGPhiMC - Centre de génétique et de physiologie moléculaire et cellulaire

We propose a data mining method to analyse protein-protein interaction (PPI) networks where biological properties are associated to the proteins. In this context, we find collections of several protein modules (set of proteins in interaction) such that all proteins in a collection share a set of biological properties. Such structure, named Collection of Homogeneous Protein Modules (CHPM), allows the analyst to get a larger picture when analysing a network of proteins. Our method is applied to a PPI network associated with gene expression data and we show that 70\% of the patterns found are significantly associated to a biological function.

Keywords: attributed graph; k-clique percolated component; graph mining; network analysis

Domains: Data Structures and Algorithms [cs.DS]; Bioinformatics [q-bio.QM]; Quantitative Methods [q-bio.QM]

Pierre-Nicolas Mougel, Christophe Rigotti, Olivier Gandrillon. Finding Collections of Protein Modules in Protein-Protein Interaction Networks. Bioinformatics and Computational Biology (BICoB), 2012, Las Vegas, United States. pp.216-222. ⟨hal-00758858⟩