Most of the work was done using protege (standalone version).
owlready2 (python package) was also used for some edition operations. See: https://pypi.org/project/Owlready2/
rdflib (python package) was used to directly generate some simple rdf files.
sparql queries, most of them executed against the inhouse JENA and Virtuoso instances. see the SPARQL folder for details.
The main imported ontologies and their useage fields are:
| Ontology | Usage | namespace (shortcut) | Prefix | TP high level entry |
|---|---|---|---|---|
| Cell Ontology | Specific cells (eg. macrophages, activated T cell) | obo | CL_ | Biological entities/cell |
| PRotein Ontology / GO | For proteins/gene (as the entries match in this specific ontology) | obo | [PR_ / GO_ | Biological entities/Cell Marker |
| UBERON | For tissue sorts (ex. 'cartilage tissue') and organs (indications) | obo | UBERON_ | Biological entities/[Excluded Region, organ part, organ, tissue] |
| OBI | For high level desription of data and their organisation. The following ontologies often refer to it | obo/IAO | [OBI_/IAO_] | many |
| CHEBI | For chemical entities (DAB, eosin, etc.) | CHEBI | CHEBI_ | material entity, chemical entity |
| EFO | extensive ontology covering most areas. Here: disease, p-values, etc. | EFO | EFO_ mostly | several |
| PATO / UO / GENEPIO / STATO | Phene descriptions and combinations | ... | [PATO_ / UO_ / GENEPIO_ / ... ] | [data item/datum label] |
The relevant terms of these ontologies were mireoted* using ontofox. The configuration files and result files are under imports.
Note also the branch refOntos in this very repository which contains most of the previously cited ontologies in a dedicated folder (not included in master as this corresponds to a significant amount of data).
In the case of EFO, two imports are generated:
- efo_import.owl the generic file including neoplasm
- efo_disease_import.owl which is 3.2mb large on its own and which contains several neoplasm sorts of tissues / diseases. efo_disease_import.owl is actually the output of tools/reorganizeImportEFODisease.py as most of the structure is lost during mireoting. An intermidiate owl (efo_disease_import_b.owl) file is used.
All proteins refer to the PRotein Ontology. This ontology has several domains. The entries used here should belong to the 'gene' category or 'external' in which case they should refer to the HGNC standardized term. This way all entries can be mapped one to one to their coding gene and which is normally identified by ex. the hugo (hgng) term. This makes the joined analysis of gene expression data and tissue phenomes easier as the same identifier can be used safely.
UO defines most units as classes. Here only the main categories (area unit, volume unit, etc.) were imported and individuals are used. Following the approach used in OBI, the individuals are given the same iri as the original class (see punning**).
The use of tpo:label allows to specify tissue phenomics specific terms.
* MIREOT: The minimum information to reference an external ontology term ** Punning: https://www.w3.org/TR/2012/REC-owl2-new-features-20121211/#F12:_Punning