Skip to main content
. Author manuscript; available in PMC: 2016 Jan 4.
Published in final edited form as: Methods Mol Biol. 2011;756:99–130. doi: 10.1007/978-1-61779-160-4_5

Fig. 3.

Fig. 3

Representation of ontological structures. Ontology of biologically relevant factors can be represented in a simple graphical structure in which parent Gene Ontology terms give rise to progeny terms (a). Parent terms are typically of a broad nature with their successive progeny possessing increasingly specific annotation (level 1 to 4). This simple graphical ontology representation though can be governed by both directed and nondirected rules. Directed ontological relationships imply a classical hierarchical parent–progeny linking between the terms, that is, parent–progeny relationships are directed downward from less complex terms to more complex terms (black arrows, panel A). However, as broad-level parent terms may lead to multiple more specific ontological terms the simple one-parent one-progeny relationship may be less likely to reflect physiological systems than the one-parent multiple-progeny ontology (b). Undirected ontological representations, however, may allow nondirected progeny to parent relationships (c). Undirected representations may lead to cyclic closed relationship loops. If, however, all of the ontological relationships are directed then it is possible to represent biological linkages into a directed acyclic graph (DAG). (d) An example of an actual DAG from input signaling data. The three major classes of ontology (GObp, GOmf, GOcc) are shown. GO term specificity increases with descent into progeny branches of the DAG. Therefore, the most statistically significantly populated ontology terms are found in the lowest areas of the DAG diagram (e.g., circled GO term groups).