System pharmacology: Application of network theory in predicting potential adverse drug reaction based on gene expression data

In drug development process, adverse drug reaction (ADR) is one of the biggest challenges to evaluate the drug safety for passing to the market. Genomic expression data following in vitro drug treatments and thus have become widely used in ADR identification and prediction. In this research, we develop the prediction method by using system pharmacology-based study. We performed the proteomic, small molecular compounds - protein interaction and ADR data based on Connectivity Map database. A major protein-drug-side effect (PDS) network and a protein-drug (PD) network were obtained and analyzed by followed network centrality study, which allows for selection of side effects that are defined as central nodes. From the result, the top ranking of novel side effects was identified. In a case study, we established prediction models for 2,3, 7, 8-tetra-chlorodibenzo-p-dioxin (TCDD) in breast cancer treatment adverse events. In conclusion, the network-based approach provided the relationship between protein targets network and side effects based on the gene expression profiles and can predict the potential side effects for new a combinatory drug in the drug development process