Data

To investigate the state and evolutionary change of biomedical ontologies, we need a comprehensive collection of ontologies. We have chosen to investigate BioPortal data since it is widely recognized as a comprehensive repository of biomedical ontologies. Parts of this and the next section summarize our previous paper with added information.

The data (Supplemental Information 1) was gathered at the following five time points: October 2012, February 2013, August 2013, December 2013 and July 2014. Currently BioPortal contains more than 400 ontologies grouped into 41 categories (e.g., Health, Anatomy, Cell). To perform the analysis, the following data had to be collected through BioPortal RESTful web services for all time points: the ontology’s full name (e.g., Gene Ontology), the ontology’s name abbreviation (e.g., GO), and the number of mappings from/to the ontology. Since the BioPortal RESTful interface changed after August 2013, we gathered the following additional data only for the first three versions of our visualizations: ontology statuses (e.g., production) and ontology versions (e.g., alpha). To analyse identified communities, we also collected number of projects and categories that community members belong to.

Our analysis depends on mapping information in BioPortal. The BioPortal web page describes mappings as:

“Mappings are associations between two or more terms in different ontologies. This association typically, but not always, represents a degree of similarity between the terms. The author of the mapping defines the semantics of a particular mapping. It is also usual for a mapping to be bi-directional, but again, this is not required. The mapping author defines directionality.”

We collected the number of all mappings between ontology pairs. The following three types of mappings are supported (please note that no information about mapping types was gathered for our current analysis):