iGMDR
iGMDR database collects predictive models of response for anti-cancer drugs. Cancer is the primary goal of precision medicine because of its heterogeneity, leading to the failure of traditional therapies. The efficacy of anti-cancer drugs usually varies from individual to individual, mainly due to different genetic backgrounds. Therefore, it is feasible to design personalized therapy strategies based on the genetic background of patients. The predictive model of drug response is a combination of genetic characteristics analyzed using big data of patient population . In recent years, with the development of pharmacogenetics research and the application of clinical sequencing of individual genome, prediction models were more and more used in clinical practices, but only a few drugs have the matching drug genetic information, and some drug genetic information was not very clear even so.
iGMDR database aims to build an open community collecting the results of the current phase of pharmacogenetics research, and extracting the prediction model of therapeutic response for anti-cancer drugs. The prediction model involves different genetic characteristics. According to different extraction techniques, it mainly includes single nucleotide variation (SNV), copy number variation (CNV), gene expression variation (GEX) and gene structure variation (SV). The database not only collects models that have already been used in clinical practices, but also the prediction models obtained in vivo or in vitro experiments. Although most of these models are not necessarily of clinical significance, the integration of these models will help us understand the mechanism of cancer therapies from a macro-perspective, understand the response mechanism of individual drugs from a micro-perspective, and further release the value of those pharmacogenetics research to promote clinical practices.
As shown in below, the iGMDR database collected 14 genetic characteristics of 1040 drugs of 144 cancer types of 30 tissue types, which formed 154,146 prediction models. The data we have integrated are the information of FDA, NCCN, ASCO and other clinical practices, and the information of pharmacogenetics experiment of model organism and cell line. In order to meet the requirements of big data analysis, we standardized information such as cancer, drugs and genetic characteristics, and stored such information using structured manner.