ElasticNet with First order Interaction of Specific Genes for drug response prediction using preclinical samples

Prediction of drug response based on the patient’s molecular profile is a challenge in the field of precision oncology, and large scale cancer genomics studies such as TCGA, GDSC, and CCLE provide molecular data for predicting drug response. Although it is still unclear whether preclinical samples can be used to predict drug response about actual patients, we used preclinical sample data to evaluate the performance of various linear and nonlinear model. Finally, we used the first order interaction of specific genes to improve prediction performance, and we used cross validation scores to verify that the algorithms we applied performed better than other algorithms.