Prediction of disease–drug relationships based on tissue specificity and direct neighbor similarity

Abstract

The pathogenesis of complex diseases is a major problem in the field of human health. The development of new drugs through traditional methods requires considerable time and money, which has not met peoples actual requirements. Recently, identifying new therapeutic effects of known drugs via drug repositioning has become an effective way to treat numerous diseases. At present, tissue-specific research has achieved some success; however, traditional drug repositioning methods rarely consider the tissue specificity of the disease. To explore the influence of tissue specificity on drug repositioning studies, this study explores the development of tissue specificity and its characteristics and proposes using direct neighbor similarity in drug repositioning based on tissue-specific data. A total of 11405 known drug–target relationships were extracted from the database DrugBank, and five cancers and their disease-causing gene data were obtained from the human Mendelian genetic database. Through the direct neighbor method and using the tissue-specific interaction network as the background network, five tissue-specific drug–disease bipartite networks were constructed, which provided potential drug–disease associations. The results were verified by the CTD (comparative toxicogenomics database) standard. The experimental results show that the accuracy of drug repositioning studies based on tissue specificity and direct neighbor measurement will provide a reliable candidate set for in vivo and in vitro experiments of new drugs, which also provides new ideas for studying drug repositioning.