KaBOB: ontology-based semantic integration of biomedical databases

Area of Science:

• Biomedical Informatics
• Data Integration
• Knowledge Representation

Background:

• Querying diverse biological databases requires a common semantic model for effective data integration.
• Existing challenges in semantic data integration hinder establishing shared identity and meaning across heterogeneous biomedical data sources.

Purpose of the Study:

• To present five processes for semantic data integration to address key challenges in biomedical data.
• To demonstrate these processes by creating the Knowledge Base Of Biomedicine (KaBOB) integrating data from 18 prominent biomedical databases.

Main Methods:

• Explicitly defining differences between biomedical concepts and database records.
• Aggregating identifiers for the same biomedical concepts across multiple data sources.
• Utilizing declarative forward-chaining rules for consistent data representation from variably sourced information.

Main Results:

• Developed and applied five semantic data integration processes solving seven key problems.
• Created KaBOB, a knowledge base integrating data from 18 biomedical databases using Open Biomedical Ontologies.
• KaBOB instance with data for humans and seven model organisms comprises approximately 500 million RDF triples; source code is open-source.

Conclusions:

• KaBOB enables querying integrated biomedical data using common concepts (genes, interactions) rather than source-specific schemas.
• KaBOB resolves common issues faced by researchers working with multi-source biomedical data.
• Provides a platform for ongoing data integration, development, and formal reasoning over integrated biomedical data.