Relations
From Program on Ontologies of Neural Structures
Contents |
Overview
This project is focused on developing formalized relations for ontologies of neuroanatomy. The main focus is on relations for recording connectivity within nervous systems: between neurons; between neurons and structural areas; between tracts and structural areas. Our goal is to create a consistently named set of inter-related, formally defined relations. The hope is that these will provide a basis for formalizing relations in ontologies generated by the neuron registry task force and the structural lexicon task force. See NRTF mappings and SL mappings for proposed mappings to this work.
Here, we outline names, descriptions and formalizations for these relations as of June 2010. This is still a work in progress and is a work in progress. We need feedback on the names, descriptions, and relations, please add your comments and suggestions. To see these, please go to Proposed Relations. If you are not familiar with OWL2, it may help to read this Primer on formalization first.
Note that the formalizations reference GO terms for neuron components. A list of such term can be found here GO neuron component terms. Feedback on problems with the definitions of these terms or requests for additional terms should be submitted to the GO tracker (note sourceforge account required).
One major limitation of the work so far is the lack of a formalization for electrical synapsing. This needs expert input from the group. A discussion page for this can be found here: A formal relation for electrical synapsing.
Criteria
Ideally, the relations (as a whole) should allow one to express a range of assertions about neuro-anatomy. However, keep the following in mind:
- It is unlikely all relations will be used in any one project (most will probably use a subset suited to their needs).
- Proposed names should be judged in terms of how clear their meaning will be to naïve users.
- No choice of name can be expected to cover all usage (but will preferably handle most).
- Chosen names should be clear, not frequently mis-interpreted.
- Names should all work for statements about singular objects e.g. neuron X is_synapsed_to neuron Y
Notes for a Reviewer
- It is anticipated that all these relations will be used with a simple quantifier pattern:
- Every X rel (relation) some Y.
- For example, if we define a relation, 'has_synapse_in', that applies between a neuron and a brain region where that neuron forms synapses, this will be used to record statements of the type: Every class X neuron has_synapses_in some region.
- For each relation, we also need to define an inverse relation.
- For has_synapse_in, we might define an inverse relation, 'has_synapse_of', that applies between a brain region and a neuron that forms synapses in it. Say we don't know whether all class X neurons have synapses in some region Y, but we know that some do, and that in every brain studied, some class X neurons can be seen to innervate region Y. We can record this by using the inverse relation of has_synapses_of and the simple quantifier pattern above: Every region Y has_synapses_of some class X neuron.
- As per OBO standard practice, underscores are in the names, but we may be able to remove them in the official version to make them more readable (this is still under discussion with OBO foundry folks).
Details of formalization strategy
- Where possible, relations are defined as an expansion to a nested class level expression in OWL that uses only core relations (from the OBO relations ontology) and references only classes from OBO Foundry ontologies.
- Expansions: The system is designed so that ontologies built with these relations are usable without expansion. The semantics of the unexpanded relations are enriched using property hierarchies and property chains. In order to avoid contradiction between these two sets of semantics (expanded and unexpanded), expansion will generate additional OWL axioms, leaving the expanded properties and their relations to each other in place.
For an brief introduction to how to read these formalizations, please see Primer on formalization.
