Skip to content

OBO modifications

Jim Balhoff edited this page Sep 28, 2017 · 13 revisions

The NCIt is modified in several ways to produce the OBO Library edition. Many of the changes aim to align with the OBO Foundry Identifier Policy. The OBO edition is produced via a Makefile.

NCIt OBO Edition

Ontology IRI

The ontology adopts a conventional OBO-style ontology IRI: http://purl.obolibrary.org/obo/ncit.owl

An OBO format version is available at http://purl.obolibrary.org/obo/ncit.obo.

The ontology version IRI incorporates the release date; multiple dated releases may have been derived from the same standard NCIt release: http://purl.obolibrary.org/obo/ncit/releases/2017-08-29/ncit.owl

The standard NCIt release version from which the OBO Edition was derived remains in the owl:versionInfo property:

<http://purl.obolibrary.org/obo/ncit.owl> owl:versionInfo "17.06d" .

Term IRIs

IRIs for classes and properties incorporate the NCIt term code into an OBO format:

  • Lymphoma: http://purl.obolibrary.org/obo/NCIT_C3208
  • Disease_Has_Associated_Anatomic_Site: http://purl.obolibrary.org/obo/NCIT_R100

Within OBO convention these IRIs can be shortened as NCIT:C3208 and NCIT:R100.

Each term is additionally linked to its NCIt code via the code annotation property (http://purl.obolibrary.org/obo/NCIT_NHC0), e.g.:

NCIT:C3208 NCIT:NHC0 "C3208" .

Annotation property substitutions

Some annotation properties are replaced by their OBO counterparts:

  • DEFINITION (NCIT:P97) => IAO:0000115
  • FULL_SYN (NCIT:P90) => oboInOwl:hasExactSynonym
  • Concept_In_Subset (NCIT:A8) => oboInOwl:inSubset

Obsoletion of role classes

NCIt includes a hierarchy of classes under NCI Thesaurus Role with the same labels as the object properties in the ontology, yet with different IRIs. The presence of these terms confuses the Protégé user interface, making it difficult to use object properties in DL queries. In the NCIt OBO Edition these terms are deprecated and relabeled by prepending "obsolete ".

Removal of multiple definitions

A small number of terms have more than one value for the definition annotation property. This violates the expectations of the OBO format serializer. To avoid problems in these cases, one of the definitions is moved to a value for rdfs:comment.

Pre-reasoning

The class hierarchy is inferred, using the ELK reasoner, and provided as asserted subclass axioms within the release file.

NCIt Plus

In addition to the above modifications, the NCIt Plus release includes more extensive axiom changes.

Ontology IRI

http://purl.obolibrary.org/obo/ncit/ncit-plus.owl

Direct integration with domain-specific ontologies

NCIt terms with mappings to concepts in other domain-specific OBO Library ontologies (e.g. cell types, cellular components, anatomy) are replaced with direct references to those terms. The NCIt term is marked as deprecated and removed from logical axioms. Currently, terms are incorporated from Uberon and CL.

Definitional axioms for the incorporated external terms are extracted from the external ontologies and incorporated into the NCIt Plus release.