UC San Diego

The past decade has seen the creation and maturation of a number of new technologies designed to study life on a genome-wide scale. However, the sheer volume of data generated by these methods surpasses the analytical and critical abilities of a single researcher. For this reason, it is necessary to create new computational methods to assist in the analysis of these new sources of data. Both yeast-two hybrid and co-immunoprecipitation followed by mass spectrometry allow the determination of binding interactions between proteins. Functional (genetic) interactions are determined via SGA (Synthetic Genetic Array) and E-MAP (Epistasis Mini-array Profile). In Chapters 2 and 3, we develop algorithms to integrate these two types of interactions together for the purpose of biological pathway discovery. Moreover, our approaches create maps of genetic interactions that provide a picture of the global organization of pathways and complexes within the cell. Expression arrays are a genome-wide quantitative assay for mRNA levels within the cell. Using fluorescent dyes, two different biological samples can be directly compared on a single array slide. In Chapter 4, we identify a gene-specific dye bias in this type of expression array data. We improve upon a maximum likelihood method in order to remove the effect of this bias. Using novel control experiments, we show that this enhanced analysis yields results that more reproducible. Complementary to expression profiling, deletion fitness profiling quantifies the relative fitness defect of every deletion strain in Saccharomyces cerevisiae under a particular stress. In Chapters 5 and 6, we discuss how to use the type of pathway information uncovered in Chapters 2 and 3 to improve the analysis of both expression and deletion fitness profiling datasets. We apply these methods to the study of two different cellular stresses in Saccharomyces cerevisiae, arsenic exposure and adaptation to oxidative stress