The Creation and Application of Bioinformatic Techniques to Improve Therapeutic Options for Cancer Patients

Tumors commonly exhibit high levels of both inter- and intra- heterogeneity. For this reason, the optimal therapeutic approach for cancer patients would ideally be a regimen that is catered toward their individual tumor. Unfortunately, rather than receiving a treatment based on the molecular profile of their tumor, most cancer patients receive a generalized therapeutic treatment based on the past performance of tumors at the same stage and location. These non-specific treatments often cause adverse side effects and may have only a slight impact on life expectancy of a patient. The limited treatment options for cancer patients are mainly because of the lack of knowledge of which prerequisites are required for a tumor to respond to an anti-cancer drug. Fortunately, with the advancement of next generation sequencing technologies, we can globally interrogate the molecular characteristics of individual tumors. Here, I present four ways to improve therapeutic options for cancer patients by utilizing and developing computational tools that analyze next generation sequencing data: 1) combining machine learning with Bayesian network structure learning to identify tissue specific mechanisms of drug response, 2) subtyping colorectal cancers to identify molecular mechanisms associated with early-onset colorectal cancer, 3) combining exome and RNA sequencing data to better identify influential tumor mutations, and 4) identifying breast cancer patients who are particularly susceptible to bone metastasis. The results from all four projects suggest that, although tumors from the same tissue typically exploit similar pathways to progress their tumor phenotype, the specific mechanisms they use to dysregulate the pathway often vary. Moreover, common genomic alterations that occur across multiple tumor types activate different tissue-specific mechanisms, which can dramatically alter the response of tumors to the same anti-cancer drug. Elucidating how tissue- and tumor-specific molecular dysregulation drives tumor phenotypes is an essential prerequisite to providing the optimal therapeutic for a cancer patient. By using the approaches and workflows I developed during my PhD, we can better suggest anti-cancer drugs that will elicit a response in cancer patients by identifying the biological pathways that are critical for tumor development in specific tissues and which molecular dysregulations may potentiate that pathway