The top-level organization of the main classes of the ontology conforms to the Basic Formal Ontology (BFO) upper ontology as shown in the following figure:
This provides a basic model for how OBI models a biomedical investigation. Although models can be very detailed, much of the design of OBI revolves round four classes: (A) ‘planned process’ (B) ‘plan specification’, (C) ‘material entity’, and (D) ‘information content entity’.
OBI’s scheme for generating linked data that describe experiments (at least down to the protocol level) is illustrated in the following figure:
Any ‘planned process’ can specify substructure through the ‘has part’ object property. An ‘investigation’ is a ‘planned process’ that has a ‘study design execution’ as its main part. A ‘study design execution’ often has parts of three main types: ‘material processing’, ‘assay’, and ‘information content entity’.
We then use the ‘has specified input’ and ‘has specified output’ object properties to link these processes to ‘material entity’ and ‘data item’ objects to describe a complete experimental scientific protocol. The actual values of data elements in OBI are mainly provided by instances of a handful of data properties, or by way of the ‘value specification’ class. We may also specify the conclusions of a study through the ‘drawing a conclusion based on data’ process which generates a ‘conclusion based on data’. Additional semantics are invoked to instantiate these models for real data, but these core classes provide the basic structure for such descriptions.