DRPO: A Deep Learning Technique for Drug Response Prediction in Oncology Cell Lines

With the invention of high-throughput screening technolo- gies, innumerable drug sensitivity data for thousands of can- cer cell lines and hundreds of compounds have been pro- duced. Computational analysis of these data has opened a new horizon in the development of novel anti-cancer drugs. Previous deep-learning approaches to predict drug sensitiv- ity showed drawbacks due to the casual integration of ge- nomic features of cell lines and compound chemical features. This paper proposes a new deep learning model, DRPO , for efficient integration of genomic and compound features in predicting the half maximal inhibitory concentrations (IC50). First, matrix factorization is used to map the drug and cell line into latent ’pharmacogenomic’ space with cell line-specific predicted drug responses. Using these drug responses, we next obtained the essential drugs using a Normalized Dis- counted Cumulative Gain (NDCG) score. Finally, the essen- tial drugs and genomic features are integrated to predict drug sensitivity using a deep model. Experimental results based on RMSE and NDCG scores on CCLE & GDSC1 datasets show that our proposed approach has outperformed the previous approaches, including DeepDSC, CaDRRes, and KMBF. With these good results, it shows great potential to use our new model to discover novel anti-cancer drugs for precision medicine.