Applying Machine Learning Methods to Suggest Network Involvement and Functionality of Genes in Saccharomyces cerevisiae

Abstract

Elucidating genetic networks provides the foundation for the development of new treatments or cures for diseased pathways, and determining novel gene functionality is critical for bringing a better understanding on how an organism functions as a whole. In this dissertation, I developed a methodology that correctly locates genes that may be involved in genetic networks with a given gene based on its location over 50% of the time or based on its description over 43% of the time. I also developed a methodology that makes it easier to predict how a gene product behaves in a cellular context by suggesting the correct Gene Ontology term over 80% of the time. The designed software provides researchers with a way to focus their search for coregulated genes which will lead to better microarray chip design and limits the list of possible functions of a gene product. This ultimately saves the researcher time and money.