Knowledge Graph Integration

The Knowledge Graph integration enables rich, interconnected data representation across the Bio-DID-Seq and BioAgents ecosystem, supporting semantic queries and intelligent data discovery.

Overview

Knowledge Graphs provide a powerful way to represent research data as interconnected entities and relationships rather than isolated documents. The Bio-DID-Seq platform leverages knowledge graphs to:

1. Enhance discoverability of research data
2. Connect related biological entities across datasets
3. Enable complex semantic queries
4. Support AI powered knowledge extraction and reasoning

Knowledge Graph Architecture

1. Core Components

RDF Triple Store

The foundation of the knowledge graph is a Resource Description Framework (RDF) triple store that represents data as subject-predicate-object triples:

# Example triple representation
<http://example.org/gene/BRCA1> <http://example.org/relation/associatedWith> <http://example.org/disease/BreastCancer> .

JSON-LD Documents

Knowledge is also represented in JSON-LD format for compatibility with web standards:

{
  "@context": {
    "bio": "http://example.org/biology/",
    "schema": "http://schema.org/"
  },
  "@id": "bio:gene/BRCA1",
  "@type": "bio:Gene",
  "bio:name": "BRCA1",
  "bio:associatedWith": {
    "@id": "bio:disease/BreastCancer",
    "@type": "bio:Disease",
    "bio:name": "Breast Cancer"
  }
}

SPARQL Endpoint

A query interface allows complex semantic queries against the knowledge graph:

PREFIX bio: <http://example.org/biology/>

SELECT ?gene ?disease WHERE {
  ?gene bio:associatedWith ?disease .
  ?disease a bio:Disease .
  ?disease bio:relatedTo bio:CancerPathway .
}

2. Integration Points

Knowledge Extraction Process

1. Document Processing

BioAgents processes research papers to extract structured knowledge:

1. Text Extraction: Convert PDF documents to machine readable text
2. Section Identification: Identify abstract, methods, results, etc.
3. Entity Recognition: Identify genes, proteins, diseases, etc.
4. Relationship Extraction: Determine connections between entities

2. Knowledge Representation

Extracted knowledge is represented in a structured format:

// Example of extracted biological relationships
interface BiologicalRelationship {
  subject: {
    term: string;
    id: string;
    type: "gene" | "protein" | "disease" | "pathway" | "compound";
  };
  // e.g., "inhibits", "activates", "associatedWith"
  predicate: string;
  object: {
    term: string;
    id: string;
    type: "gene" | "protein" | "disease" | "pathway" | "compound";
  };
  evidence: {
    text: string;
    source: string;
    confidence: number;
  };
}

3. Knowledge Graph Integration

Extracted knowledge is integrated into the knowledge graph:

BioAgents Knowledge Graph Features

1. AI-Assisted Knowledge Enrichment

BioAgents enhances the knowledge graph through:

• Context aware entity disambiguation: Resolving ambiguous entity references
• Missing relationship inference: Inferring implicit connections
• Cross Cross Platform DID Resolutionreference resolution: Linking to external knowledge bases

2. Decentralized Knowledge Graph (DKG)

Integration with OriginTrail DKG provides:

export const dkgInsert: Action = {
  name: "INSERT_MEMORY_ACTION",
  description: "Create a memory on the OriginTrail DKG",
  handler: async (runtime, message, state, options, callback) => {
    // DKG integration code
    const createAssetResult = await DkgClient.asset.create(
      { public: knowledgeAsset },
      { epochsNum: 12 }
    );
    
    return true;
  }
};

Bio-DID-Seq Knowledge Graph Integration

1. DID-Based Knowledge Access

Each knowledge graph node can be associated with a DID:

{
  "@context": [
    "https://www.w3.org/ns/did/v1",
    "https://w3id.org/biodata/v1"
  ],
  "id": "did:bio:123456789abcdefghi",
  "service": [
    {
      "id": "did:bio:123456789abcdefghi#kg",
      "type": "KnowledgeGraphService",
      "serviceEndpoint": "https://sparql.bio-did-seq.example/endpoint"
    }
  ],
  "knowledgeGraph": {
    "rootCID": "QmXg9Pp2ytZ14xgK35M6iTC2Vz6jR9zYgooNp2UHPTMnPN",
    "entityCount": 1457,
    "relationshipCount": 3892,
    "lastUpdated": "2025-06-17T14:23:45Z"
  }
}

2. UCAN-Based Authorization

Access to knowledge graph data is controlled through UCAN capabilities:

{
  "header": {
    "alg": "EdDSA",
    "typ": "JWT"
  },
  "payload": {
    "iss": "did:key:z6MkhaXgBZDvotDkL5257faiztiGiC2QtKLGpbnnEGta2doK",
    "aud": "did:key:z6MkgYGF3tLg4MzGCLjkpkzYZqhrMkWGBMPXATFksYp1PxJh",
    "nbf": 1617228000,
    "exp": 1617314400,
    "att": [
      {
        "with": {"scheme": "did", "hierPart": "bio:123456789abcdefghi", "parameters": {}},
        "can": "kg/query"
      }
    ]
  },
  "signature": "..."
}

Query Capabilities

1. Basic SPARQL Queries

Users can query the knowledge graph using SPARQL:

PREFIX bio: <http://bio-ontology.org/>

SELECT ?gene ?diseaseName WHERE {
  ?gene bio:associatedWith ?disease .
  ?disease a bio:Disease ;
           bio:name ?diseaseName .
  FILTER(CONTAINS(?diseaseName, "Alzheimer"))
}

2. Natural Language Queries

BioAgents supports natural language queries that are translated to SPARQL:

public async query_agents(query: string): Promise<(string, Vec<string>)> {
    // Natural language processing to SPARQL conversion
    // Query processing logic
    return (answer, sources);
}

Example query: "What genes are associated with Alzheimer's disease that also interact with the APOE pathway?"

Use Cases

1. Knowledge Discovery

Researchers can discover hidden connections between biological entities and can enable delegation for specific Collaborators:

2. Hypothesis Generation

BioAgents can generate testable hypotheses based on knowledge graph patterns:

1. Identify gaps in the knowledge graph
2. Apply reasoning to suggest potential connections
3. Rank hypotheses by plausibility
4. Provide supporting evidence

3. Data Integration

The knowledge graph connects data across multiple sources:

• Research papers
• Experimental results
• External databases
• Clinical data
• Genomic information

Future Developments

Upcoming enhancements to the knowledge graph integration include:

1. Federated Queries: Query across multiple knowledge graphs
2. Temporal Knowledge Representation: Track how knowledge evolves over time
3. Uncertainty Representation: Express confidence levels in relationships
4. Reasoning Capabilities: Advanced inference across the knowledge graph
5. Multi Modal Knowledge: Represent non textual data (images, etc.)