Network analysis in pharmacology for drug repositioning and drug target identification

Abstract

Network biology represents and analyzes biological systems as complex networks consisting of various kinds of cellular molecules and relationships between them to understand cellular functions at system level. Viewing biological systems via the network concept may also help us improve drug discovery by revealing more complex aspect of drug action in cellular networks. To apply network analysis into drug discovery process, first of all, we gathered comprehensive knowledge from diverse disciplines related to drug discovery and integrated it into a pharmacological tripartite network database (PharmDB) consisting of drugs, targets, diseases and their relationships. Secondly, we developed shared neighborhood scoring algorithm, a new method proper to analyze our pharmacological tripartite network since existing network analysis approaches use various projection methods to convert a multi-partite network into several mono-partite networks causing information loss. By combining PharmDB with the shared neighborhood scoring algorithm, we can explore the pharmacological tripartite network to identify new drug targets, to design drug repositioning and new drug combination, and to predict potential side effect of a drug. Thirdly, in addition to network topology analysis based on prior knowledge, we tried network inference, a kind of data-driven approach to utilize large-scale gene expression profiles into network analysis. We constructed a transcriptional network using ARACNE as a network inference algorithm and analyzed it to extract master regulators or transcription factors (TFs) for known prognostic signature genes of colorectal cancer. All of this works represents that network biology can be a fully integrated solution to better, more efficient drug discovery and development.