Measurement of drug-drug similarity by identifying gene correlation module signature from drug response profile

Abstract

Drug repositioning is re-evaluating approved drugs or experimental drugs to find their novel indications. This repositioning based drug discovery can save total costs and time required for the drug development and significantly reduce risks occurred by safety problems. To support the novel drug repositioning candidates, various prediction methods have been developed including side-effect similarity, chemical similarity, literature mining, drug-target association prediction, drug-disease association prediction, etc. In addition to these methods, large-scale drug response profiles and a differential expression pattern comparison method, which can be used to compare the profiles, were introduced by Connectivity Map. Major application of expression similarity search methods was searching large-scale gene expression experiment repositories with content based manner. After the advent of Connectivity Map data, the application of expression similarity search is extended to the identification of novel drug-drug and drug-disease associations in gene transcriptional level. Various methods have been made to support the analysis of the drug response profile. However, there are not enough studies focused on what kinds of gene signature can represent the drug response profile and systematic performance comparison of different similarity measurement methods. In this research for the prediction of novel drug-drug similarity, we made comprehensive re-sources for the pharmaceutical research. After construction of the background resources, we identified gene correlation modules from profiles treated with same drug and used the modules to analyze the drug-drug differential expression similarity. First, we constructed a background resource for the drug discovery and development. First of all, we collected disease associated genes and drug information from various resources and integrate them into a standardized disease term set. The drug response profiles were collected from the Co...