Department of Medical Informatics and Clinical Epidemiology
Oregon Health & Science University
In the study of rare and undiagnosed diseases it is of critical importance to identify potential gene candidates for those disease in order to establish an appropriate treatment protocol. However, there may be insufficient resources available to investigate rare and undiagnosed diseases, so it may be necessary to identify potential gene candidates from patient clinical observations to narrow the focus of the investigation. Two approaches for identifying gene candidates are investigated and compared: the OWLSim package for comparing the semantic similarity of phenotypes and the Phenolog approach for identifying orthologous phenotypes by searching for orthologous gene enrichment between phenotypes.
In comparing the two approaches by measuring ranked-recall of known disease causative genes, the Phenolog approach performed better than the OWLSim package. However, each approach was able to exclusively identify known disease causative genes, indicating that a combined approach would provide a better retrieval of correct gene candidates than either method alone.
School of Medicine
Laraway, Bryan, "Comparative analysis of semantic similarity and gene orthology tools for identification of gene candidates for human diseases" (2015). Scholar Archive. 3741.