Drug Repositioning Research Based On Module Distance And Multi-source Random Walk

Drug repositioning, which means discovering new indications for marketed drugs, is undoubtedly a very effective solution to drug development problem since new drug research and development is difficult, costly and time-consuming now. The main purpose of drug repositioning is to find potential drug-disease relation based on calculation and integrated genetic and other biological data. A series of measures like literature mining, biological pathway enrichment analysis can be used to verify predictions. Currently, most drugs repositioning methods take all diseases into account which lack of specific evaluation and thus causing an inaccurate forecasting results. For these reasons, this paper proposes two research programs:(1) Firstly, narrow the range of diseases study to cancer. Then, predict the potential drugs related to these cancer diseases based on a complex network module distance calculation method. Finally, the overall predicting results are analyzed to be relatively reliable but literature mining and enrichment of functional path analysis in this method are not satisfactory. It may due to lack of integrity on data. Based on this, we propose the second research program.(2) Firstly, the aim of the disease study is determined to be liver cancer. Secondly, based on multivariate random walk, calculate the similarity between liver cancer related gene sets and other drug related gene sets. Then, establish a drugs-drugs network based on the similarity result and pick out drugs directly connected to already known liver cancer drugs, and grade them. A higher score indicates the drug is more likely to be a potential treatment for liver cancer. Finally, the high score drugs were analyzed and found to be accurate and reliable whether based on literature mining, Comparative Toxicogenomics Database(CTD) validation or function pathway enrichment analysis. The result further illustrates that focusing on one disease and using an improved random walk algorithm help to achieve the goal of drug repositioning, the key of drug repositioning is object-specific and efficient computing method.