Date of Award
Doctor of Philosophy (PhD)
Genomics & Computational Biology
Virus-host interactions are being cataloged at an increasing rate using protein interaction assays and small interfering RNA screens for host factors necessary for infection. These interactions can be viewed as a network, where genes or proteins are nodes, and edges correspond to associations between them. Virus-host interac- tion networks will eventually support the study and treatment of infection, but first require more data and better analysis techniques. This dissertation targets these goals with three aims. The first aim tackles the lack of data by providing a method for the computational prediction of virus-host protein interactions. We show that HIV-human protein interactions can be predicted using documented human peptide motifs found to be conserved on HIV proteins from different subtypes. We find that human proteins predicted to bind to HIV proteins are enriched in both documented HIV targeted proteins and pathways known to be utilized by HIV. The second aim seeks to improve peptide motif annotation on virus proteins, starting with the dock- ing site for protein kinases ERK1 and ERK2, which phosphorylate HIV proteins during infection. We find that the docking site motif, in spite of being suggestive of phosphorylation, is not present on all HIV subtypes for some HIV proteins, and we provide evidence that two variations of the docking site motif could explain phos- phorylation. In the third aim, we analyze virus-host networks and build on the observation that viruses target host hub proteins. We show that of the two hub types, date and party, HIV and influenza virus proteins prefer to interact with the latter. The methods presented here for prediction and motif refinement, as well as the analysis of virus targeted hubs, provide a useful set of tools and hypotheses for the study of virus-host interactions.
Evans, James P., "Modeling Virus-Host Networks" (2010). Publicly Accessible Penn Dissertations. 260.