Computational analysis of cis-regulatory mechanisms underlying cellular response to drug perturbations

In this thesis, I, alongside my varied and talented collaborators, aim to integrate two seemingly disparate subfields of biology: pharmacogenomics and regulatory genomics. The former concentrates on how inherited factors such as SNPs and genes inform the response to medical interventions designed to ameliorate disease states, processes, and phenotypes; ontologically, it resides at the intersection of genomics and pharmacology. Regulatory genomics, however, directs its attention fully to the biological mechanisms by which genes express themselves, without regard to application or context. As such, it is viewed within the intersection of genomics and systems biology. A systems perspective is relevant because last 20 years of regulatory genomics research have concluded that (a) genes do not necessarily express binarily as a (b) function of logical operations on the presence/absence of a (c) finite set of protein regulators binding to the (d) genes immediate upstream DNA sequence, but rather, that the span of regulators ranges from protein DNA binding factors to other such nuclear factors (including the DNA itself) up and through various signaling molecule networks to the various components of the cell, terminating at the interface of the cell and extracellular environment the membrane. DNA accessibility and conformation, cellular type, micro-RNAs are all real factors that influence gene expression. This thesis, however, assumes the more nave assumptions of gene regulation namely that the DNA binding proteins, called Transcription Factors (TFs), influence the kinetics of the transcriptional process by binding in certain configurations proximal to the genes transcription start site. In combining both pharmacogenomics and regulatory genomics, it was necessary to make simplifications and assumptions to the biology on both sides to make the problem tractable; however, as a result, we were able to construct statistical and probabilistic models that capture novel biological associations that would otherwise remain unknown. Namely, associations between nuclear regulators of genes and drug response. In doing so, we provide a novel link in connecting these two disparate views of biology into a more cohesive and singular picture