Computational approach for cancer drug repositioning based on the transcriptional perturbation analysis

Abstract

Even though the level of investment in drug research and development (R&D) has increased dramatically, the success rate remains relatively low. As an alternative, drug repositioning, an identification of new indications for existing drugs, is providing a viable solution to overcome this low productivity problem. Currently, most of network-based drug repositioning methods only focus on the relationship between drug target genes and known cancer-related genes. However, from recent studies, it has been revealed that transcription factors related to cell development and cell proliferation play key roles in cancer development. In addition, most of currently prescribed anti-cancer drugs either directly target the cancer-related transcription factors or indirectly regulate them via signaling cascade. So, regarding to this cancer-related characteristic of transcription factors, if we can find drugs that could directly or indirectly affect transcription factors, much accurate anti-cancer drug repositioning would be possible. In this study, we developed a novel network-based cancer drug repositioning method to identify new indications of existing drugs for cancer treatment. We inferred drugs having direct or indirect effects on cancer-related transcription factors, by using random walk with restart algorithm on tissue/cancer-specific biological networks. We applied our method to three case studies: acute myeloid leukemia, prostate cancer and non-small cell lung cancer. In addition, for each case study, we carried out the same inference process for three conventional approaches using known cancer-related genes, differentially expressed cancer-related genes and transcription factors among the cancer-related genes, to compare the prediction performance to our method. In all three cases, our method showed better prediction of known anti-cancer drugs than conventional approaches. In addition, we found that many top-ranked drugs which are not included in the answer drugs have literature evidences. From this result, we could confirm the validity of our hypothesis

the importance of transcriptional factor in cancer treatment. In recent studies, it has been found that other human diseases, like type-2 diabetes and auto immune disease, are also associated with mutations in transcription factors. We believe that our method can be applied to those diverse diseases and recommend viable drugs from drug repositioning.