Dramatically decreased cost of sequencing has enabled us to sequence the genomes of hundreds of individuals with various diseases. Yet, pinpointing the gene variants responsible for the development of heterogeneous diseases remains a particularly hard task because the same phenotypic outcome (disease) can result from a myriad of combinations of different alterations across the genome. In this talk, I will describe two projects that further our ability to computationally highlight potential disease causing genes by examining the disease genomes in the context of biological networks.    

First, I will present a novel network-based approach to uncover cancer-relevant genes from large-scale cancer genomics data. My approach tackles cancer mutational heterogeneity by utilizing per-individual mutational profiles. My method is based on the expectation that if a pathway is relevant for cancer, then (1) many individuals will have a somatic mutation within one of the genes comprising the pathway and (2) the genes comprising the pathway will interact with each other and together form a small connected subcomponent within the larger network. I will provide an intuitive formulation relying on balancing the size of the connected subgraph with covering many patients. I will describe a machine learning-like schema for selecting the value of the single required parameter and both an integer linear programming framework and a fast heuristic for optimizing the objective function. I will demonstrate the method’s outstanding ability to identify cancer-relevant genes, especially those mutated at very low rates.

Second, I will describe a general computational framework that uses prior knowledge of disease associated genes to guide a network-based search for novel ones based on newly acquired information. While many methods that directly consider either the prior knowledge or the newly acquired information to uncover disease-relevant genes have been developed, I develop a new network-based approach that combines these two distinct types of information. I will show how such a framework built upon graph kernels and guided random walks has the power to uncover new disease associated genes by applying it to different heterogeneous diseases, including macular degeneration and eight cancer types.