Identification of Topological Network Modules in Perturbed Protein Interaction Networks

Mihaela E Sardiu,Michael P Washburn,Laurence Florens,Brad Groppe,Joshua M Gilmore

doi:10.1038/srep43845

Abstract

Biological networks consist of functional modules, however detecting and characterizing such modules in networks remains challenging. Perturbing networks is one strategy for identifying modules. Here we used an advanced mathematical approach named topological data analysis (TDA) to interrogate two perturbed networks. In one, we disrupted the S. cerevisiae INO80 protein interaction network by isolating complexes after protein complex components were deleted from the genome. In the second, we reanalyzed previously published data demonstrating the disruption of the human Sin3 network with a histone deacetylase inhibitor. Here we show that disrupted networks contained topological network modules (TNMs) with shared properties that mapped onto distinct locations in networks. We define TMNs as proteins that occupy close network positions depending on their coordinates in a topological space. TNMs provide new insight into networks by capturing proteins from different categories including proteins within a complex, proteins with shared biological functions, and proteins disrupted across networks.

Highlights

To begin our investigation into the topology of perturbed protein interaction networks, we first carried out a quantitative proteomic analysis of the S. cerevisiae INO80 protein interaction network
Replicates were performed in our analysis which resulted in a total of 31 purifications (Supplementary Table S1A)
Taken together these results suggest that protein complexes, pathways and protein interactions between these modules tend to be unstable in response to perturbation of the INO80 protein interaction network

Summary

Introduction

We have used TDA to study the conservation of human and yeast chromatin remodeling networks[18] and the associations of the uncharacterized WDR76 protein with DNA damage and chromatin remodeling proteins[19] In this body of work, we investigated the capabilities of TDA for the analysis of perturbed protein interaction networks from two different species. These modules could contain proteins with different features like proteins in a complex, proteins with distinct biological functions, or proteins altered by the system. Capturing these classes of proteins would typically require multiple different computational approaches asking a specific question. We term these modules Topological Network Modules (TNMs), which are proteins that occupy close network positions depending on their coordinates in a topological space

Methods

Results

Conclusion