Cargando…

Ion Channel ElectroPhysiology Ontology (ICEPO) – a case study of text mining assisted ontology development

BACKGROUND: Computational modeling of biological cascades is of great interest to quantitative biologists. Biomedical text has been a rich source for quantitative information. Gathering quantitative parameters and values from biomedical text is one significant challenge in the early steps of computa...

Descripción completa

Detalles Bibliográficos
Autores principales: Elayavilli, Ravikumar Komandur, Liu, Hongfang
Formato: Online Artículo Texto
Lenguaje:English
Publicado: American Medical Informatics Association 2016
Materias:
Acceso en línea:https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5001744/
https://www.ncbi.nlm.nih.gov/pubmed/27570648
Descripción
Sumario:BACKGROUND: Computational modeling of biological cascades is of great interest to quantitative biologists. Biomedical text has been a rich source for quantitative information. Gathering quantitative parameters and values from biomedical text is one significant challenge in the early steps of computational modeling as it involves huge manual effort. While automatically extracting such quantitative information from bio-medical text may offer some relief, lack of ontological representation for a subdomain serves as impedance in normalizing textual extractions to a standard representation. This may render textual extractions less meaningful to the domain experts. METHODS: In this work, we propose a rule-based approach to automatically extract relations involving quantitative data from biomedical text describing ion channel electrophysiology. We further translated the quantitative assertions extracted through text mining to a formal representation that may help in constructing ontology for ion channel events using a rule based approach. We have developed Ion Channel ElectroPhysiology Ontology (ICEPO) by integrating the information represented in closely related ontologies such as, Cell Physiology Ontology (CPO), and Cardiac Electro Physiology Ontology (CPEO) and the knowledge provided by domain experts. RESULTS: The rule-based system achieved an overall F-measure of 68.93% in extracting the quantitative data assertions system on an independently annotated blind data set. We further made an initial attempt in formalizing the quantitative data assertions extracted from the biomedical text into a formal representation that offers potential to facilitate the integration of text mining into ontological workflow, a novel aspect of this study. CONCLUSIONS: This work is a case study where we created a platform that provides formal interaction between ontology development and text mining. We have achieved partial success in extracting quantitative assertions from the biomedical text and formalizing them in ontological framework. Availability: The ICEPO ontology is available for download at http://openbionlp.org/mutd/supplementarydata/ICEPO/ICEPO.owl