Identifying cancer-related genes via network feature learning and multi-omics data integration

The primary purpose of this project is to identify cancer-related genes using PPI networks and machine learning and data representation models. We integrated the protein-protein interaction (PPI) network with gene expression data in this analysis. We obtained useful features using gene expression data for each gene and then used different feature learning algorithms, such as DeepWalk, node2vec, and GCN, to find a representation for each node (gene) in the PPI network. The extracted embedding vectors were used to predict cancer-related genes using machine learning models, such as SVM and deep neural networks.