Identification of cancer drivers, conserved alteration patterns and evolutionary trajectories in tumors

Over the past decade, high-throughput sequencing efforts have provided an unprecedented opportunity to identify genomic alterations that can lead to changes in gene regulation, protein structure, and function. During tumor progression, cancer cells accumulate a multitude of genomic alterations, a small fraction of which provide tumor cells with selective advantage - known as "driver" alterations. However, most of them are inconsequential "passenger" alterations that are effectively neutral and greatly outnumber driver alterations. Moreover, due to a high amount of heterogeneity, alteration landscapes of tumors differ between different patients and different sites. This thesis first presents HIT'nDRIVE, a method that integrates genomic and transcriptomic data to identify a set of patient-specific, sequence-altered potential driver genes, with sufficient collective influence over dysregulated transcripts through interactome. Applied to 2200 tumors, HIT'nDRIVE revealed many potentially clinically actionable driver genes and demonstrated its robustness in selecting cancer-implicated drivers. The results also show that small network modules seeded by HIT'nDRIVE-selected drivers significantly improve classification of cancer phenotypes and drug efficacy in pan-cancer cell lines compared to alternative methods and approaches. Next, a method for detection of functionally meaningful and recurrent alteration patterns within gene interaction networks, cd-CAP, is presented. In a number of TCGA data sets, cd-CAP identified large subnetworks with identically conserved alteration patterns (across many tumor samples), that were significantly associated with patients' clinical outcome. As multi-region, time-series and single cell sequencing data become more widely available, computational methods have been developed with the goal of inferring the subclonal composition and evolutionary history of tumors from tumor biopsy sequencing data. Unfortunately, the phylogenetic trees reported for many tumor samples differ significantly from other tumors with similar characteristics. This thesis presents CONETT, the first computational method for detection of conserved trajectories of alteration events in tumor evolution. Applied to two multi-region sequencing data sets of 100 tumors each, CONETT confirms all findings of the original studies and identifies additional repeated trajectories.