Identifying Drug Interactions And Apply Them To Predict Cancer Drug Sensitivity On Cell Lines

With the development of high-throughput sequencing,we now have access to a large amount of biological data,and data on genes and anticancer drugs and protein-protein interaction networks are constantly being updated and enriched."In the face of biological big data,how to establish a mathematical model for rapid processing of large data and effective analysis to maximize the hidden information found in the important information" is an important research topic in the field of bio-mathematics.Studying and predicting the sensitivity of anticancer drugs through mathematical models is a basic work in cancer research and an important method to predict the sensitivity of anticancer drugs.It is also of practical significance to genomics and medicine.It is well-known that one of the lasting challenges of precision medicine is to choose the right anticancer drug for each individual patient.Patients testing drugs in large clinical trials are a way of assessing their efficacy and toxicity,but it is impractical to test hundreds of drugs currently under development.Therefore,a preclinical prediction model is highly desirable because it enables the drug to respond in parallel to predictions of hundreds of cell lines.The current methods of predicting drug susceptibility using gene expression data are based on the assumption that genes and genes are independent of each other and that they do not consider the interaction between genes.We call this prediction method a one-dimensional anti-cancer drug sensitivity prediction.We constructed three methods to study the gene-gene interaction and applied it to the prediction of drug sensitivity,which is called the prediction of the sensitivity of binary anti-cancer drugs.These three methods are DGCorNet,DGRNet and DGPPINet respectively.The three methods are based on the prediction of the sensitivity of a single anti-cancer drug.The first method is mainly based on the correlation coefficient between genes and genes.The second chapter mainly uses the linear regression between genes and genes.Finally,One approach is mainly the application of protein-protein interactions.On the one hand,based on the gene expression data and the activity area data of anticancer drugs,this paper studies the interaction between genes and genes under the action of drugs and proposes a new idea for drug sensitivity prediction.On the other hand,we use the matrix filling algorithm to establish the drug prediction model.Here we use the elastic mesh regression algorithm to calculate the predictive value of anti-cancer drug sensitivity.