Graph Neural Networks for Drug Response Prediction
Jiehuang Zhu, Xianfeng Liu, Jin Zhang, Sheng Yi Wu · 2023
Drug response prediction plays a crucial role in precision medicine, such as cancer analysis and treatment. Due to the uncertainty of drug efficacy and the heterogeneity of cancer, predicting drug response in vitro is expected to assist in guiding anticancer drug design and understanding cancer biology. Based on the Genomics of Drug Sensitivity in Cancer (GDSC) and the Cancer Cell Line Encyclopedia (CCLE), we propose a deep learning method for predicting drug sensitivity. Specifically, we first construct graphs of drugs and cell lines in the benchmark dataset. And we design a drug feature extraction module based on Graph Transformer and a cell line feature extraction module based on Graph Attention Networks (GAT) to capture the embeddings of drugs and cell lines. Finally, we feed the drug embeddings and cell line embeddings into the module for prediction. Experiment results show that our method outperforms several advanced models in drug response prediction, demonstrating the good performance and enormous application potential in precision medicine. In addition, the results of ablation experiments demonstrate the effectiveness of the modules in our model.