To link and integrate different Open Biological and Biomedical Ontologies (OBO) library ontologies, we generated an OLOBO ontology that uses the Basic Formal Ontology (BFO) as the upper level ontology to align, link, and integrate all existing 9 formal OBO Foundry ontologies. Our study identified 524 ontology terms shared by at least two ontologies, which form the core part of OLOBO. Current OLOBO includes 978 ontology classes, 13 object properties, and 90 annotation properties. OLOBO was queried, analyzed, and applied to identify potential deprecated terms and discrepancy among different OBO library ontologies.
2.1
Instead of using all >150 ontologies, we chose the official 9 OBO Foundry ontologies (Table 1). OBO Foundry website includes their default ontology download websites. We downloaded the OWL source code of these ontologies.
We developed an in-house Java program that uses OWLAPI
to process ontology OWL files. Many ontologies also reuse
other ontology terms. To integrate all the ontologies, we
first identified and collected all ontology terms shared by at
least two ontologies. This step identified the minimal sets of
ontology terms shared by different ontologies. To notify
from which ontology a term is imported, we generated a
new annotation property called ‘term from ontology’
(OLOBO_0000001). Next, we used Ontofox
The OLOBO source code is openly available at GitHub:
https://github.com/biomedontology/olobo. OLOBO is
deposited in Ontobee: http://www.ontobee.org/ontology/OLOBO.
Ontobee SPARQL queries was used to query OLOBO
OLOBO was used to identify deprecated terms, and we used the Ontobeep tool (http://www.ontobee.org/ontobeep) to compare with OLOBO and other OBO ontologies to identify potential discrepancy among ontologies.
# Prefix
Ontology full name # of terms # of shared terms 1 BFO 2 ChEBI 3 DOID 4 GO 5 OBI 6 PATO 7 PO 8 PRO 9 XAO
Our study found 524 classes, object properties, and annotation properties shared by at least two ontologies. In total, our pipeline removed 178,625 terms from the original 9 OBO ontologies. The OLOBO statistics is available on Ontobee: http://www.ontobee.org/ontostat/OLOBO. OLOBO includes 978 classes, and 103 object or annotation properties.
The overall OLOBO structure (Fig. 1) is aligned with BFO (http://ifomis.uni-saarland.de/bfo/), the recommended upper-level ontology in the OBO Foundry, and >100 ontologies are aligned to BFO. The chosen 9 ontologies are all aligned with BFO as defined by the ontologies themselves or manually by us.
UBERON - PO, XAO, ZFA
GO ChEBI
ChEBI (covering PR terms)
OBI (covering NCBITaxon terms) covering PATO terms
DOID
GO
GO
UBERON - PO, XAO, ZFA 3.2
Our study identified three deprecated PATO terms, PATO_0001237 (quality of a single physical entity), PATO_0001238 (quality of related physical entities), and PATO_0001631 (relational spatial quality), which are still used in OBI. PATO has recommended new PATO terms to replace these obsolete terms, e.g., ‘physical object quality’ (PATO_0001241) to replace PATO_0001238. We thus submitted a track issue to OBI (https://github.com/obiontology/obi/issues/826) to indicate such an issue.
Another application was to use OLOBO for ontology comparison and discrepancy detection. Ontobeep is an Ontobee-based ontology alignment and comparison tool. Our usage of Ontobeep identified that the UBERON term ‘anatomical entity’, which are included in OLOBO and OBI, also exists in CLO but under a different parent term. Such an ontology alignment and surveillance study is expected to identify achievements as well problematic issues in the topdown alignment study. These will significantly support collaborative ontology development and reuse.
The basic idea of our project is to generate an integrated upper level application ontology that links OBO ontologies in an integrated format. This study only aligned existing 9 official OBO ontologies. We plan to extend OLOBO to include the other candidate OBO ontologies.
The challenge of ontology integration and orthogonality
is real. Using ontology term labels to analyze OBO ontology
orthogonality
In addition to the demonstrated OLOBO usages, OLOBO may be used in other cases. For example, to develop a new OBO-oriented ontology, we may use OLOBO as the starting point to facilitate the ontology development process and lead to more efficient alignment. The usage of OLOBO may also provide an integrated ontology system for efficient software development for features such as consistency checking and standard integration. Given the large number of terms among different ontologies, it is practically infeasible to develop a single ontology that covers all ontology terms from all possible ontologies. However, OLOBO offers an integrative framework to link and integrate different ontologies in a semantically consistent strategy.
This work was partially supported by a BD2K-LINCS DCIC external data science research award, which is part of the grant U54HL127624 (BD2K LINCS Data Coordination and Integration Center, DCIC) awarded by the National Heart, Lung, and Blood Institute through funds provided by the trans-NIH LINCS Program and the trans-NIH Big Data to Knowledge (BD2K) initiative (http://www.bd2k.nih.gov). LINCS is an NIH Common Fund projects.