Aber-OWL: a framework for ontology-based data access in biology

Many ontologies have been developed in biology and these ontologies increasingly contain large volumes of formalized knowledge commonly expressed in the Web Ontology Language (OWL). Computational access to the knowledge contained within these ontologies relies on the use of automated reasoning. We have developed the Aber-OWL infrastructure that provides reasoning services for bio-ontologies. Aber-OWL consists of an ontology repository, a set of web services and web interfaces that enable ontology-based semantic access to biological data and literature. Aber-OWL is freely available at http://aber-owl.net.

💡 Research Summary

**
AberOWL is an integrated infrastructure that brings automated reasoning to the large collection of biomedical ontologies expressed in OWL. It first gathers ontologies from user‑specified URIs using the OWL API, extracts class labels and textual definitions according to OBO Foundry conventions, and stores the labels in a trie for fast autocomplete and lookup. Each ontology is then classified with the ELK reasoner, which implements the OWL EL profile and provides polynomial‑time computation of subclass, superclass, and equivalent class hierarchies even for ontologies containing thousands of classes.

Users submit queries in Manchester OWL Syntax via a web interface or RESTful API. The system translates the query string into an OWL class expression, runs ELK to retrieve the relevant class hierarchy, and returns the results as a JSON array containing the ontology document URI, class IRI, label, and definition.

Two main services exploit these reasoning results. AberOWL : PubMed builds a Lucene index of MEDLINE/PubMed abstracts, titles, and full‑text articles from PubMed Central. For a given OWL query, the service concatenates the labels of all retrieved classes with a logical OR and searches the Lucene index, thereby returning literature that mentions not only the explicit query term but also any inferred subclasses or related concepts. This enables truly semantic literature retrieval, such as finding articles about “tetralogy of Fallot” when the query is “ventricular septal defect” because the ontology infers the former as a subclass.

AberOWL : SPARQL extends SPARQL 1.1 queries by embedding OWL query results directly into VALUES or FILTER clauses. A user can request all GO terms that are “part of some apoptotic process,” obtain the corresponding class IRIs, and bind them to a SPARQL variable. The variable is then used to filter data in external Linked Data endpoints such as UniProt or EBI BioModels. The service also handles URI scheme mismatches by converting OBO‑style IRIs into prefix forms required by specific endpoints, ensuring seamless integration across heterogeneous SPARQL services.

The platform is open‑source (GitHub) and modular: while the default deployment uses ELK, users may replace it with HermiT, Pellet, or other OWL reasoners via the OWL API. The repository currently hosts the full set of OBO ontologies plus user‑requested additions, and the web interface allows immediate querying, PubMed search, or SPARQL expansion without additional configuration.

In summary, AberOWL demonstrates a practical realization of the ontology‑based data access paradigm for biology. By coupling ontology metadata, fast EL‑profile reasoning, and web‑service integration, it enables detection of inconsistencies, enrichment of incomplete annotations, and uniform querying across literature and Linked Data sources. This capability represents a significant advance over existing ontology portals that provide only static metadata, positioning AberOWL as a valuable tool for researchers seeking semantic, inference‑driven access to biomedical knowledge.