Repositioning Drugs Based On Molecular Network

Drug discovery is a long and expensive procedure, where it typically takes about twelve years and around two billion dollars to bring a new drug to market. Despite the increasing research spending on drug development, the new drug approval rate has a tendency to reduce. Therefore, it is necessary to develop new approaches for drug discovery. In recent years, the emerging of different types of omics data has provided new opportunities for this problem. Recently, identifying new indications for old drugs, i.e. drug repositioning, attracts a lot of attention, which can significantly reduced the cost and risk of developing new drugs due to the guarantee of toxicity and therapeutic effects of existing drugs.In this thesis, some computational approaches were proposed to repurpose drugs for breast cancer based on molecular networks. With these approaches, some drugs that were not originally used for breast cancer were repositioned for the disease. The thesis is mainly composed of the following three parts:1) Identification of drug targets based on protein domains. With the assumption that drug-protein interactions are determined by drug-domain interactions, a novel approach was presented to predict the interactions between drugs and proteins. Based on the predicted drug-domain interactions, a drug association network was constructed. The drugs with similar therapeutic effects were found to be grouped into same module(s), implying the drugs interacting with the same domain tend to have similar therapeutic effects, thereby validating the effectiveness of the proposed approach to some extent. The discovery of new targets for known drugs can help identify new indications of the drugs.2) Drug repositioning for breast cancer based on molecular network. In theory, an effective drug is able to reverse the expression of the disease signature. With this assumption, a new approach was presented to identify new indications for known drugs based on the disease signatures detected from molecule interactions network, where the signature were regarded as the downstream pathways affected by the causal genetic mutations of the disease. The approach was applied to repositioning drugs for breast cancer, and some of the predictions were validated by evidences from literature.3) Identification of genes related to Aristolochic acid(AA) pathogenesis. With different types of omics data generated under AA treatment, a regulatory network consists of transcriptional regulations and post-transcriptional regulations was constructed. Some network modules were detected to be enriched with differentially expressed genes by AA treatment. Further analysis indicates that these modules were enriched in pathways and biological processes related to AA induced nephrotoxicity and carcinogenicity, implying the genes in those modules may be related to AA pathogenesis. The disease gene modules identified here can help screen drugs that can be used to treat the disease, and are therefore useful for repositioning drugs for the disease.