JMIR Publications

ABSTRACT

Background:

Alzheimer’s Disease (AD) is a devastating neurodegenerative disease, of which the pathophysiology is insufficiently understood, and the curative drugs are long-awaited to be developed. Computational drug repurposing introduces a promising complementary strategy of drug discovery, which benefits from an accelerated development process and decreased failure rate. However, generating new hypotheses in AD drug repurposing requires multi-dimensional and multi-disciplinary data integration and connection, posing a great challenge in the era of big data. By integrating data with computable semantics, ontologies could infer unknown relationships through automated reasoning and fulfill an essential role in supporting computational drug repurposing.

Objective:

The study aimed to systematically design a robust Drug Repurposing-Oriented Alzheimer’s Disease Ontology (DROADO), which could model fundamental elements and their relationships involved in AD drug repurposing and integrate their up-to-date research advance comprehensively.

Methods:

We devised a core knowledge model of computational AD drug repurposing, based on both pre-genomic and post-genomic research paradigms. The model centered on the possible AD pathophysiology and abstracted the essential elements and their relationships. We adopted a hybrid strategy to populate the ontology (classes and properties), including importing from well-curated databases, extracting from high-quality papers and reusing the existing ontologies. We also leveraged n-ary relations and nanopublication graphs to enrich the object relations, making the knowledge stored in the ontology more powerful in supporting computational processing. The initially built ontology was evaluated by a semiotic-driven and web-based tool Ontokeeper.

Results:

The current version of DROADO was composed of 1,021 classes, 23 object properties and 3,207 axioms, depicting a fundamental network related to computational neuroscience concepts and relationships. Assessment using semiotic evaluation metrics by OntoKeeper indicated sufficient preliminary quality (semantics, usefulness and community-consensus) of the ontology.

Conclusions:

As an in-depth knowledge base, DROADO would be promising in enabling computational algorithms to realize supervised mining from multi-source data, and ultimately, facilitating the discovery of novel AD drug targets and the realization of AD drug repurposing.