MHGCN: A Multi-Channel Hybrid Graph Convolutional Neural Network for Cancer Drug Response Prediction
Abstract:
Due to the heterogeneity of cancer cells, personalized treatment plans for cancer patients remain a continuous concern. High-throughput drug screening technology has led to the development of deep learning models that generate personalized therapies. However, most existing models fail to account for the topological relationships between cell line-drug pair (CDP) nodes, thereby ignoring their intrinsic connections. This paper proposes a multi-channel hybrid graph convolutional neural network (MHGCN) for predicting cancer drug response (CDR). First, we define CDPs by integrating gene expression and drug molecular fingerprints. These CDPs are refined through denoising autoencoders to eliminate noise. Second, we compute pairwise cosine similarities among CDPs to build a similarity network, while simultaneously establishing a heterogeneous response graph connecting cell lines and drugs. Third, MHGCN processes the CDP network via graph convolutional layers and generates the response matrix through linear projection. Concurrently, a heterogeneous graph convolutional neural network learns the response heterogeneous network. Following data augmentation, we derive feature embeddings for cell lines and drugs, then compute their similarity matrix. Finally, CDR predictions are generated through weighted matrix fusion of these components. To the best of our knowledge, MHGCN represents the first framework explicitly incorporating CDP topology into CDR prediction. Experiments demonstrate MHGCN's statistically significant improvements over state-of-the-art methods.
Insights
This study introduces a novel deep learning model, the multi-channel hybrid graph convolutional neural network (MHGCN), to improve personalized cancer drug response prediction by considering cell line-drug pair (CDP) topology. MHGCN significantly enhances prediction accuracy over existing methods.
Area of Science:
- Computational biology
- Genomics
- Pharmacology
Background:
- Personalized cancer treatment is challenging due to cancer cell heterogeneity.
- Deep learning models aid personalized therapy but often ignore intrinsic relationships in cell line-drug pair (CDP) data.
- Existing models lack the integration of topological information for improved cancer drug response (CDR) prediction.
Purpose of the Study:
- To propose a novel multi-channel hybrid graph convolutional neural network (MHGCN) for accurate cancer drug response (CDR) prediction.
- To incorporate the topological relationships of cell line-drug pairs (CDPs) into a deep learning framework.
- To enhance personalized therapy by improving the prediction of drug efficacy in cancer patients.
Main Methods:
- Defined CDPs by integrating gene expression and drug molecular fingerprints, refined using denoising autoencoders.
- Constructed a CDP similarity network and a heterogeneous response graph connecting cell lines and drugs.
- Employed MHGCN with graph convolutional layers and a heterogeneous graph convolutional neural network for feature embedding and response prediction, followed by weighted matrix fusion.
Main Results:
- The proposed MHGCN framework explicitly incorporates CDP topology, a novel approach for CDR prediction.
- MHGCN demonstrated statistically significant improvements in prediction accuracy compared to state-of-the-art methods.
- The model effectively integrates multi-modal data and graph structures for robust CDR prediction.
Conclusions:
- MHGCN offers a significant advancement in predicting cancer drug response by leveraging topological information.
- The framework provides a more accurate and personalized approach to cancer treatment planning.
- This study highlights the importance of considering network topology in deep learning models for biological data analysis.
