Cross-sectional evaluation of disease classes in public ontologies

How do public biomedical ontologies model disease? This dataset evaluates 278 classes that ontologies in BioPortal, the Ontology Lookup Service and the OBO Foundry offer as disease. For each class it records four things: what kind of entity the class takes a disease to be (continuant, occurrent or both); whether it keeps the disease apart from its classification; whether it draws the line between healthy and diseased as crisp or graded; what, in its view, makes a condition pathological. It also rates the quality of each class’s definition, comments and superclass. 146 classes met the inclusion criteria; the other 132 are kept together with the reason for their exclusion. The evaluation was carried out for the master’s thesis “Eine umfassende Ontologie der Wunde”. It provides the basis for the Mid Layer Disease Ontology (MILD, https://doi.org/10.5281/zenodo.23863289) and the Wound Ontology (WONT, https://doi.org/10.5281/zenodo.23864083). The package contains: data/: the master table, one row per evaluated class; both ontologies are generated from it. ontology/: the evidence ontology with all 278 records and the source definitions, comments and superclasses they rest on; the synthesis ontology with the 146 included classes; the declarations of the annotation properties; 156 modules holding each class’s superclasses as its own ontology states them; and a catalog that resolves all imports locally for Protégé and ROBOT. figures/: the selection flow diagram and the summary of the evidence (PNG and SVG). code/: the Python scripts for search, extraction, consistency checks, export and figures. docs/: the review protocol, the methods, the quality assessment workflow and the plan for an independent second-reviewer validation. Data, ontologies, figures and documents are licensed under CC BY 4.0, the code under the MIT License. Statements quoted from the evaluated ontologies keep the rights of their sources.