Index > Neuron Lang overview Edit on GitHub

Neuron Lang overview

Neuron Lang overview

The Neuron Phenotype Ontology

We are creating a new neuron ontology. This ontology is built on the idea that neuron types are fundamentally bags of phenotypes.

Phenotypes

In our model phenotypes are the outputs of some measurement or set of measurements and computations/classifications. Ideally the process for documenting exactly how each phenotype was measured would be included as well, but that is currently outside the scope of this project. The current major categories of phenotypes (with example identifiers are):

  • species [NCBITaxon:10116]
  • strain [???]
  • developmental stage [uberon developmental stage?]
  • brain region [UBERON:0000955]
  • morphology [ILX:]
  • electrophysiology [ILX:]
  • expression [NCBIGene:, PR:000004967; CHEBI:18243]
  • projection pattern [ILX:]
  • cell type specific connectivity [ILX:]
  • circuit role [ILX:]

Phenotype Predicates

The core of our neuron ontology is based around a set of phenotype predicates that define relationships between phenotypes an neurons. Some examples of these predicates (relationships) are has morphological phenotype, has electrophysiological phenotype, and has expression phenotype. In the first two cases these predicates correspond directly to phenotype classes that are in ttl/NIF-Neuron-Phenotype.ttl because there are not other ontologies that have provided a consistent delineation for many of these types. In the case of expression phenotypes there are a number of different sources of identifiers which all imply that slightly different measurements have been made. For example a CHEBI identifier would suggest that a stain against a particular small molecule might have been used, whereas the use of a NCBIGene identifier would imply that RNAseq had been used to detect the expression of a gene transcript. While these combinations can be used to provide some implicit information about the original type of measurement, all predicate-phenotype pairs are really proxies for the experimental protocol that was used and how the resulting measurement maps onto the categorical names provided. In the future we will provide a way to link these phenotype assertions to more detailed information about the exact experimental protocol used.

Bagging

All neurons are modelled using owl:equivalentClass and owl:complementOf statements on owl:Restriction s. We have also developed a python library to regularize the bagging process. By using owl:equivalentClass and owl:disjointWith or owl:complementOf we are able to model both positive and negative phenotypes, so for example a parvalbumin positive (PV+) and somatostatin negative neuron (SOM-) would translate to the following. Using complement of works in this context because of the use of owl:intersectionOf to construct the bag.

SOMEID:1234567 a owl:Class ;
    owl:equivalentClass [ a owl:Class ;
            owl:intersectionOf (
                    NIFCELL:sao1417703748
                    [ a owl:Restriction ;
                        owl:onProperty ilx:hasExpressionPhenotype ;
                        owl:someValuesFrom NIFMOL:nifext_6 ]
                    [ a owl:Class ;
                        owl:complementOf [ a owl:Restriction ;
                                owl:onProperty ilx:hasExpressionPhenotype ;
                                owl:someValuesFrom PR:000015665 ] ) ] .

Species specific identifiers

One of the advantages of using a phenotype based approach is that when there are species specific genes or proteins it is possible to be completely explicit about the exact gene or gene product that was measure/detected. When we then construct a high level neuron such as a PV+ neuron, we can then leverage species general identifiers to make it explicit that we are making assertions about phenotypes that are evolutionarily conserved.

WARNING

At the moment if you see an ILX: identifier in one of the ontology files that is involved in modelling neurons it is temporary and should not be used.

Protege development setup

  1. Follow the ontology development setup doc. Be sure not to miss the section on setting up the catalog file.
  2. Make sure you are working on the neurons branch (run git checkout neurons).
  3. Open ttl/npo.ttl from disk in protege.
  4. Follow these sections of the ontology processes document as you edit the file.

Setup and Example notebook

Notes:

Date: 2022-12-21T16:23:01-05:00

Author: Tom Gillespie

Created: 2022-12-22 Thu 01:37

Validate