DualGCN: a dual graph convolutional network model to predict cancer drug response

Abstract

Background Drug resistance is a critical obstacle in cancer therapy. Discovering cancer drug response is important to improve anti-cancer drug treatment and guide anti-cancer drug design. Abundant genomic and drug response resources of cancer cell lines provide unprecedented opportunities for such study. However, cancer cell lines cannot fully reflect heterogeneous tumor microenvironments. Transferring knowledge studied from in vitro cell lines to single-cell and clinical data will be a promising direction to better understand drug resistance. Most current studies include single nucleotide variants (SNV) as features and focus on improving predictive ability of cancer drug response on cell lines. However, obtaining accurate SNVs from clinical tumor samples and single-cell data is not reliable. This makes it difficult to generalize such SNV-based models to clinical tumor data or single-cell level studies in the future. Results We present a new method, DualGCN, a unified Dual Graph Convolutional Network model to predict cancer drug response. DualGCN encodes both chemical structures of drugs and omics data of biological samples using graph convolutional networks. Then the two embeddings are fed into a multilayer perceptron to predict drug response. DualGCN incorporates prior knowledge on cancer-related genes and protein–protein interactions, and outperforms most state-of-the-art methods while avoiding using large-scale SNV data. Conclusions The proposed method outperforms most state-of-the-art methods in predicting cancer drug response without the use of large-scale SNV data. These favorable results indicate its potential to be extended to clinical and single-cell tumor samples and advancements in precision medicine.

Deep learning methods for drug response prediction in cancer: Predominant and emerging trends

“…Integrating multiomics into the learning process further exacerbates the already problematic feature size of single omics data. A few papers indeed demonstrate significant performance boost with multiomics ( 45 , 49 , 108 ) but the majority report only marginal improvement ( 104 , 108 , 118 , 126 , 141 , 152 ). As opposed to DREAM participators which utilized agreed-upon datasets and scoring metrics, the DRP models in Supplementary Table 1 substantially differ among them as discussed earlier, largely contributing to discrepancies and mixed conclusions.…”

“…Analysis across multiple models and omics types is required to evaluate the predictive capabilities of individual data types and their subsets and make unbiased and coherent conclusions. Such analysis should incorporate recent trends which encode biological information such as protein-protein interactions (PPI), gene correlations, and pathway information ( 45 , 49 , 109 , 111 ).…”

“…Due to substantial progress in applying GNNs to drug discovery and development ( 94 , 95 ), it was relatively straightforward to adopt similar techniques to DRP. Papers propose models that convert drug SMILES into graphs and utilize modern GNN layers to encode latent drug representations ( 49 , 103 – 106 , 114 ). The common approach is constructing a unique graph per drug with nodes and edges representing atoms and bonds, respectively, where node features are the properties of each atom.…”

“…Since a primary goal is to make DRP models useful in improving patient care, the biological domains can be further categorized into clinical (human patient data) and pre-clinical (cell lines, etc.). Considering this distinction, models can exploit data in different ways: train and test on preclinical ( 44 – 46 ), train and test on clinical, train on preclinical and test on clinical ( 47 – 49 ), and models that leverage both preclinical and clinical data during the training process and then test on clinical ( 50 – 54 ). Models that use data from mixed domains are generally more challenging to develop because they require extra steps in data preparation, advanced modeling techniques, and robust performance evaluation analysis (Section 6.2.2).…”

Omics Data and Data Representations for Deep Learning-Based Predictive Modeling

“…A GCN took the drug data and predicted the response across a variety of tumors. In [ 145 ], a GCN learned a graph of genes that were related to cancer and PPIs, and it was trained with drug chemical structures and multi-omics data of cancer cells, and it learned to predict the drug response, thus it surpassed most of the existing methods.…”