Constructing a Graph Database for Semantic Literature-Based Discovery.

Area of Science:

• Biomedical Informatics
• Computational Biology
• Knowledge Discovery

Background:

• Literature-based discovery (LBD) generates novel hypotheses by synthesizing information from existing scientific texts.
• Biomedical knowledge is often represented as networks of concepts and their relationships.
• Existing databases like SemMedDB, extracted using SemRep from Medline, store these semantic relations.

Purpose of the Study:

• To evaluate the suitability of graph databases for literature-based discovery.
• To implement LBD algorithms using a graph database and compare with relational approaches.
• To explore efficient methods for managing and querying large-scale biomedical semantic networks.

Main Methods:

• Transformed the SemMedDB relational database into the Neo4j graph database.
• Implemented core LBD discovery algorithms using the Cypher query language.
• Compared the performance and conceptual simplicity of graph databases against traditional SQL for network data.

Main Results:

• Storing semantic biomedical data is more intuitive in a graph database structure.
• Implementing LBD algorithms in Cypher is conceptually simpler than using SQL.
• Graph databases offer advantages for representing and querying complex biomedical networks.

Conclusions:

• Graph databases provide a more natural and efficient environment for semantic literature-based discovery.
• Cypher query language simplifies the implementation of LBD algorithms compared to SQL.
• This approach enhances the potential for discovering new biomedical relationships.