Searching the overlap between network modules with specific betweeness (S2B) and its application to cross-disease analysis

Marina L Garcia-Vaquero,Margarida Gama-Carvalho,Javier De Las Rivas,Francisco R Pinto

doi:10.1038/s41598-018-29990-7

Abstract

Discovering disease-associated genes (DG) is strategic for understanding pathological mechanisms. DGs form modules in protein interaction networks and diseases with common phenotypes share more DGs or have more closely interacting DGs. This prompted the development of Specific Betweenness (S2B) to find genes associated with two related diseases. S2B prioritizes genes frequently and specifically present in shortest paths linking two disease modules. Top S2B scores identified genes in the overlap of artificial network modules more than 80% of the times, even with incomplete or noisy knowledge. Applied to Amyotrophic Lateral Sclerosis and Spinal Muscular Atrophy, S2B candidates were enriched in biological processes previously associated with motor neuron degeneration. Some S2B candidates closely interacted in network cliques, suggesting common molecular mechanisms for the two diseases. S2B is a valuable tool for DG prediction, bringing new insights into pathological mechanisms. More generally, S2B can be applied to infer the overlap between other types of network modules, such as functional modules or context-specific subnetworks. An R package implementing S2B is publicly available at https://github.com/frpinto/S2B.

Highlights

Disruption of a gene sequence may cause the dysfunction of the encoded protein, which can trigger the onset of a disease
This paper shows that S2B predicts cross-disease genes, providing new insights into the molecular mechanisms of Motor Neuron degenerative Diseases (MND)
To evaluate the potential of S2B, we focused on the Motor Neuron Diseases (MND) Amyotrophic Lateral Sclerosis (ALS) and Spinal Muscular Atrophy (SMA)

Summary

Introduction

Disruption of a gene sequence may cause the dysfunction of the encoded protein, which can trigger the onset of a disease. Network-based DG prioritization methods aim to recover complete disease modules, using network interactions of known DGs to predict new DG candidates. One such method, DIAMOnD4, starts from the set of known www.nature.com/scientificreports/. Diseases sharing phenotypes exhibit alterations in similar functional pathways, and their disease modules are more likely to overlap[5,15] Based on this similarity, researchers have identified common functions among the network neighbors of genes associated with Alzheimer’s and Parkinson’s diseases[16], and looked for common neighbors of proteins associated with autism spectrum disorders[17]

Objectives

Methods

Results

Conclusion