Deshpande, Raamesh2013-12-182013-12-182013-11https://hdl.handle.net/11299/161629University of Minnesota Ph.D. dissertation. Major: Computer Science. Advisor: Chad L. Myers. 1 computer file (PDF); xv, 163, appendix A.Model organisms are often used as a test-bed for the development of new genomic technologies and computational approaches. For example, the yeast Saccharomyces cerevisiae was the first eukaryote to have its entire genome sequenced, paving the way for the sequencing of the human genome. Beyond genome sequencing, yeast and other model organisms have been extensively used for reverse genetics technology development. Reverse genetics is a general approach for studying biology where the genome is perturbed in precise ways (e.g. targeted gene deletion), to gain functional information about the perturbed genes from the resulting phenotypic changes. With developments of new genomic technologies, reverse genetics at a genome-wide scale has become a reality. This dissertation focuses on the development of several computational methods for scaling up the reverse genetics experiments in model organisms as well as for exploring the generated genomics data with the ultimate goal of understanding and translating these data for use in applications for human therapeutics. One such method I developed is COMPRESS-GI which compresses the deletion collection by 95% such that the compressed set still remains highly informative for drug discovery analyses. This compression is critical for conducting chemical genomics experiments on natural products available in extremely limited quantities. I also conducted a systematic comparison of different profile similarity measures for genetic interaction networks which was crucial in discovering dot product as one of the most robust similarity measure. Enabled by these methods, we have conducted chemical genomics experiments for more than 10,000 natural products in yeast and now aim to discover therapeutically interesting compounds for human diseases. For this problem and a more general problem of translating and comparing genomic data across species, we developed a computational method neXus. Furthermore, we have started working on applications that could benefit from discovery of large number of drug-targets. One application is discovery of cancer targets using synthetic lethal interactions; however, very few synthetic lethal interactions are known in human so we developed a novel approach of discovering cancer relevant synthetic lethal interactions by translating the wealth of genetic interactions in model organisms to human.en-USActive subnetworksCancerDrug discoveryGenetic interactionsProfile similaritySynthetic lethalityThesis or Dissertation