Drug Response Prediction Using XGBOOST

Abstract

Abstract: An essential issue in computational personalised medicine is the prediction of drug responses. There have been several proposals for approaches to this problem that rely on machine learning, particularly deep learning. Nevertheless, these approaches often portray the medications as strings, an implausible representation of molecules. Furthermore, there has been a lack of comprehensive consideration of interpretation, such as whether mutations or copy number aberrations contribute to the medication response. Graph DRP, a new approach based on graph convolution networks, is suggested as a solution to the issue in this research. Cell lines were displayed as double vectors of genetic abnormalities in Graph DRP, whereas medications were shown as sub-atomic chartsthat straightforwardly caught the bonds among particles.