An ontological analysis of some biological ontologies

INTRODUCTION The functional importance of biological entities makes their understanding, analysis, and representation essential in modern biology. Arguably, semantic representation necessary for machine interoperability is a far more difficult task than syntactic representation, necessitating conceptual schema and ontologies for in-silico biological knowledge representation. Biological ontologies are increasingly being developed for prediction, big data integration in semantic web, visualization, unstructured data interpretation, annotation, and eHealth ontology. Despite being widely used, deficiencies exist (Kumar and Smith, 2003; Kumar et al., 2004; Mougin and Bodenreider, 2005; Pal, 2006; Schulz, 2006) in their concepts, relations, and frameworks in general, leading to difficulties in semantic interoperability and integration, and possibility of wrong prediction after using them. In this opinion article, I attempted for the first time (in my knowledge) to show that some characteristic inadequacies of biological ontologies could be detected and prevented by using the philosophically inspired OntoClean method (Guarino, 2002) and the top-level DOLCE ontology (Masolo et al., 2009), both of which have well-founded formal semantics, and finally proposed an outline of a novel ontology framework which aims to remove existing deficiencies. Though preliminary, my arguments suggest that it would be worthy to look deeper into the use of OntoClean and DOLCE toward detecting ontological inadequacies and improving them, a detailed analysis of which is left as a future work. I may state that, this discussion is not meant to criticize any of the ontologies, but to present some arguments on their respective design choices when seen in the light of OntoClean and DOLCE.