Cell-Type Specific Gene Interaction Networks Using Bayesian Integration to Predict Kidney Disease Relevant Genes

Abstract

Kidney diseases are responsible for more annual deaths than many cancers, yet the underlying molecular pathways leading to their progression are still not fully understood. Studying the genes and pathways involved in kidney diseases is vital for understanding their progression and for identifying potential therapeutic targets. Leveraging diverse genome scale data, this thesis builds functional networks to represent the interaction between gene pairs for different cell types in the kidney. The functional networks are built using Bayesian integration which combines datasets based on the strength of their evidence and relevance to the system of interest. With the networks, a classifier learns the connectivity patterns of disease relevant genes and predicts additional relevant genes. The top performing classifiers and their predicted genes are further evaluated through functional enrichment and module detection to identify associated pathways. Through this analysis, the most promising kidney disease relevant genes were identified. The most promising genes include some which have recently been independently found to be implicated in the associated disease or involved in pathways known to be involved in disease progression. Further research and targeted genome wide association studies are needed to further elucidate the roles of the predicted genes in their associated kidney disease.