Font Size: a A A

Incorporation of Knowledge for Network-based Candidate Gene Prioritization

Posted on:2013-07-03Degree:Ph.DType:Dissertation
University:University of PittsburghCandidate:Kimmel, ChadFull Text:PDF
GTID:1459390008475116Subject:Computer Science
Abstract/Summary:
In order to identify the genes associated with a given disease, a number of different high-throughput techniques are available such as gene expression profiles. However, these high-throughput approaches often result in hundreds of different candidate genes, and it is thus very difficult for biomedical researchers to narrow their focus to a few candidate genes when studying a given disease. In order to assist in this challenge, a process called gene prioritization can be utilized. Gene prioritization is the process of identifying and ranking new genes as being associated with a given disease. Candidate genes which rank high are deemed more likely to be associated with the disease than those that rank low. This dissertation focuses on a specific kind of gene prioritization method called network-based gene prioritization. Network-based methods utilize a biological network such as a protein-protein interaction network to rank the candidate genes. In a biological network, a node represents a protein (or gene), and a link represents a biological relationship between two proteins such as a physical interaction.;The purpose of this dissertation was to investigate if the incorporation of biological knowledge into the network-based gene prioritization process can provide a significant benefit. The biological knowledge consisted of a variety of information about a given gene including gene ontology (GO) functional terms, MEDLINE articles, gene co-expression measurements, and protein domains to name just a few. The biological knowledge was incorporated into the network's links and nodes as link and node knowledge respectively. An example of link knowledge is the degree of functional similarity between two proteins, and an example of node knowledge is the number of GO terms associated with a given protein. Since there were no existing network-based inference algorithms which could incorporate node knowledge, I developed a new network-based inference algorithm to incorporate both link and node knowledge called the Knowledge Network Gene Prioritization (KNGP) algorithm.;The results showed that the incorporation of biological knowledge via link and node knowledge can provide a significant benefit for network-based gene prioritization. The KNGP algorithm was utilized to combine the link and node knowledge.
Keywords/Search Tags:Gene, Network-based, Node knowledge, Candidate, Given disease, Biological knowledge, Incorporation, Associated
Related items