|Ph.D Student||Kveler Ksenya|
|Subject||Text Mining of a Computational Model of the|
Inter-Cellular Immune Network to Reval Novel
|Department||Department of Medicine||Supervisor||Professor Shai Shen-Orr|
|Full Thesis text|
Protective immunity draws on functionality elicited by complex communication of numerous cell types. Though these complex signaling interactions are highly studied and many principles of individual cell behavior have been identified, the information is not readily accessible, resulting in a focus on a limited set of consensus knowledge. This problem is intensified by the recent entrance of multi-dimensional measurement technologies. High system complexity, together with tremendous volume of accumulated knowledge, challenge the human capability of reasoning over immune data. Complicated inter-cellular relations and whole system effects are difficult to capture or understand. To address the deluge of knowledge and make it accessible for system level reasoning, we built immuneXpresso, a Text Mining engine, that structures and standardizes knowledge of immune inter-cellular communication. We applied it to PubMed to identify directional relations between 340 cell types and 140 molecules (cytokines and chemokines) across thousands of diseases, establishing a first-of-a-kind large-scale map of immune signaling. Leveraging the breadth of this network, we predicted and experimentally verified novel cell-cytokine interactions, as well as built a global immune-centric view of diseases whose architecture we used to predict new cytokine-disease associations. This standardized knowledgebase (www.immunexpresso.org) paves the way for rationalized interpretation of high-dimensional immune data, transforming immunology to a systematic, model-based science.