Omics-Driven Cancer Driver Gene Prediction with Enhanced Graph Convolutional Networks
Cancer remains one of the most complex and multi-factorial diseases, characterized by the accumulation of genomic aberrations such as mutations, copy number variations, and epigenetic alterations. Research in several different domains is focused on finding the molecular drivers behind this disease. This remains a critical challenge in cancer research that requires comprehensive analysis of genomic data. The ability of deep learning approaches to handle a large amount of data shows the tremendous potential for analyzing genomic data. Several approaches have been used for understanding the underlying causes of cancer through deep learning. In particular, graph convolutional networks (GCNs) have emerged as a powerful tool for identifying cancer driver genes. GCNs operate on graph-structured data, an ideal representation for genomic data as it captures the complex functional relationships between genes. This thesis proposes a novel approach for identifying cancer driver genes that employ GCNs to leverage graph-based features of multi-omics data. Different inherent graph-based features of genomic data are fed into GCNs, which are trained in a semi-supervised manner. Subsequently, helping us with the classification of genes as cancer driver genes or passenger genes. We demonstrate the effectiveness of our approach by evaluating it on a combination of multi-omics data. Our approach outperforms existing state-of-the-art methods in terms of precision and recall on different protein interaction networks. Overall, our study illustrates the superiority and effectiveness of GCNs for detecting cancer-driver genes and contributes to our understanding of the underlying molecular mechanisms of cancer. The insights gained from this research will pave the way for the development of targeted therapies and more effective cancer treatments.
M. Farooq, "Omics-Driven Cancer Driver Gene Prediction with Enhanced Graph Convolutional Networks", M.S. Thesis, Machine Learning, MBZUAI, Abu Dhabi, UAE, 2023.