-
Views
-
Cite
Cite
David S DeLuca, Elena Beisswanger, Udo Hahn, Rainer Blasczyk, An automatically curated MHC ontology for intelligent immunogenetic information retrieval (B42), The Journal of Immunology, Volume 178, Issue 1_Supplement, April 2007, Page LB9, https://doi.org/10.4049/jimmunol.178.Supp.B42
Close - Share Icon Share
Abstract
Accurate information retrieval of scientific articles is complicated by the intricacies of natural language. Ontologies are useful for formalizing scientific concepts and providing a foundation to mediate between formal, computationally accessible, representations and informal natural language expressions. The terminology associated with the MHC exemplifies the need for a sophisticated translation process between a natural language query and the relevant documents. For example, a researcher searching for “peptides that bind MHC” should be presented with a text containing “peptides were eluted from HLA-A2.” To achieve this goal, we have developed a multi-species MHC ontology which encompasses all human MHC alleles and serologic groups. Because of the rapid rate of new allele discovery, allele incorporation is automated using the IMGT/HLA and IPD-MHC databases. In the context of the StemNet knowledge management project, this MHC ontology contributes to machine-learning-driven text annotation, as well as to search-string interpretation and mapping. In this way, StemNet aims to offer researchers a platform to efficiently access the goldmine of clinically relevant immunogenetic information, which is otherwise inaccessible by standard search engines and databases.
This work was financed by the Federal Ministry of Education and Research, Germany, funding number 01DS001C; the authors are solely responsible for content.
