Identifying disease modules and components of viral infections based on multi-layer networks

Abstract

With the emergence of multi-dimensional data, we can more comprehensively analyze the pathogenic mechanisms of complex diseases, and thereby improve the diagnosis, treatment and prevention of these diseases. This study presents a novel multi-layer network-based strategy that integrates multi-dimensional data, and identifies disease-related modules and components of viral infections. We first propose a systematic method that constructs a virus-host interaction network with three layers: a viral protein layer, a host protein layer and a host gene layer. This method combines the data of high-throughput gene expression, viral protein interactions, virus-host interactions, protein-protein interactions and transcriptional regulatory relationships. To extract the underlying mechanisms of viral infections from the multi-layer networks, we identify the intra-layer and cross-layer modules, and investigate the conserved modules across multiple datasets. The essential components in the multi-layer networks are detected by singular-value decomposition. The identified conserved modules and essential components are combined into a functional enrichment analysis that reveals their contributions to influenza virus replication. By this analysis, we elucidate the common and specific mechanisms of the replication cycles of two influenza strains. By combining the different layers of information, we can comprehensively understand pathogenic mechanisms of complex diseases.