Ziyan Yang

Mentors: Jia Tao, Fadi Towfic

Computer Science Department


Reproducing the Disease Ontology Based on Human Phenotypes Classifications


Generally, diseases are classified based on similar phenotypes. For example, diabetes type I and diabetes type II are both classified as a type of diabetes based on inability of patients to produce insulin causing high blood sugar over long periods of time. This classification method is based on the symptoms (phenotypes) that are associated with disease. Currently the methods for clustering diseases is largely manual and, as a result, slow and error-prone. One way to automate the disease clustering is to utilize existing resources such as the Human Phenotype Ontology (HPO, http://www.human-phenotype-ontology.org/contao/index.php/downloads.html) that lists symptoms associated with human diseases. Based on the data from HPO, we will produce an automated, repeatable pipeline to classify human diseases based on their phenotypes. Our second goal is then to compare our automated approach against current manual approaches based on the Human Disease Ontology (http://disease-ontology.org/). This research will produce a quick, automated means to classify new diseases based on our existing knowledge of their symptoms and to help identify potential treatments to disease based on existing treatments for other diseases with similar symptoms.