Machine learning in cancer genomics

Abstract

Cancer is caused by an increased amount of cell growth in an area due to alterations in protein synthesis caused by mutations in certain genes known as cancer driver genes. Determining the gene regulatory network for the genes involved in the pathway can enable us to identify the driver genes that are responsible for cancer so that drugs targeting such genes can be developed. Advancements in DNA microarray technologies have made time series gene expression level data available for further analysis to infer the underlying gene regulatory network. My goal in this project is to infer gene regulatory networks from time series gene expression datasets using machine learning approaches, with a focus on recurrent neural networks and iteratively reweighted least squares with decorrelation. This report covers the various research papers that I studied, the approaches that I took and the results that I obtained for the task of inferring the gene regulatory network from time series gene expression data.