SPOKE

SPOKE integrates heterogeneous biomedical data into a semantically typed knowledge graph of 27 million nodes and 53 million edges to support precision-medicine research and discovery.

Key Features:

• Knowledge graph composition: A large-scale KG comprising 27 million nodes across 21 node types and 53 million edges across 55 edge types aggregated from 41 external databases.
• Ontology framework: Built on a framework of 11 ontologies that provide structural integrity, precise mappings, and semantic typing across the graph.
• Automated update pipeline: Python scripts automate weekly download, verification for integrity and completeness, and creation of 'parent tables' for nodes and edges.
• Semantic relationships: Encodes semantically meaningful relationships to connect biomedical concepts and enable precise mappings between entities.
• REST API: Provides programmatic access that translates complex graph queries into queryable endpoints for downstream analysis.
• Neighborhood exploration: Enables exploration of local graph neighborhoods to examine relationships and context around entities.

Scientific Applications:

• Hypothesis generation: Supports generation of mechanistic or associative hypotheses by linking disparate biomedical entities via semantic edges.
• Data exploration: Facilitates exploratory analysis of large heterogeneous biomedical datasets through graph-based relationships and neighborhoods.
• Precision medicine strategies: Provides integrated contextual knowledge to support development and evaluation of personalized medicine approaches.

Methodology:

Construction uses 11 ontologies and Python scripts that perform weekly data downloads, integrity and completeness checks, and create 'parent tables' for nodes and edges; a REST API exposes graph query functionality.