Skip to main content

​
TuringDB Example Notebooks

This collection of notebooks demonstrates how to use TuringDB for real-world analytical examples. Each notebook focuses on a different domain and use case, from fraud detection to biological graph exploration, leveraging the performance and versioning capabilities of TuringDB. Explore the full set of notebooks on GitHub: github.com/turing-db/turingdb-examples

​
Domain: Finance & Fraud Detection

​
Paysim Financial Fraud Detection

Uses the synthetic PaySim dataset to simulate a transactional graph. Highlights how graph queries can uncover suspicious transaction patterns and how TuringDB supports fast multi-hop detection for fraud rings and mule accounts. Query examples: 
# Find all transactions between two accounts
MATCH (a1:Account)-->(t:Transaction)-->(a2:Account)
RETURN a1.id, a1.type, t.amount, t.type, a2.id, a2.type, t.is_fraud

# Find all 2-hops transaction chains
MATCH (a1:Account)-[s1:SENT]->(t1:Transaction)-[r1:RECEIVED]->
      (a2:Account)-[s2:SENT]->(t2:Transaction)-[r2:RECEIVED]->
      (a3:Account)
RETURN a1.id, a2.id, a3.id, t1.id, t1.type, t1.amount, t1.step, t1.is_fraud ,
       t2.id, t2.type, t2.amount, t2.step, t2.is_fraud

​
Crypto Orbitaal Fraud Detection

Demonstrates risk analysis on crypto transaction data. Focuses on exploring wallets, exchanges, and fund flows to detect potential illicit activity through graph-based features such as suspicious loops and layered structures. Query examples: 
# Find all 4-hops transaction chains
MATCH (a:Entity)-[:TRANSACTION]->(b:Entity)-[:TRANSACTION]->
      (c:Entity)-[:TRANSACTION]->(d:Entity)
RETURN a, b, c, d

​
Domain: Transport & Logistics

​
London Transport (TfL)

Models the London transport network as a graph. Investigates connections between stations, line transfers, and potential applications for routing or infrastructure planning using fast traversal queries. Query examples: 
# Check if there is a path between
# Notting Hill Gate and High Street Kensington stations
# in 1 hop specifically using Circle Line
MATCH (start {displayName: 'Notting Hill Gate'})-[e:CONNECTED {Line: 'Circle'}]->
      (end {displayName: 'High Street Kensington'})
RETURN start.displayName, e.Line, end.displayName

# Find 8-hops paths between Paddington and Blackfriars stations
MATCH (start:Station{displayName:"Paddington"})-[e1:CONNECTED]->
      (s1:Station)-[e2:CONNECTED]->
      (s2:Station)-[e3:CONNECTED]->
      (s3:Station)-[e4:CONNECTED]->
      (s4:Station)-[e5:CONNECTED]->
      (s5:Station)-[e6:CONNECTED]->
      (s6:Station)-[e7:CONNECTED]->
      (s7:Station)-[e8:CONNECTED]->
      (end:Station{displayName="Blackfriars"})
RETURN start.displayName, start.Note, e1.Line,
       s1.displayName, s1.Note, e2.Line,
       s2.displayName, s2.Note, e3.Line, 
       s3.displayName, s3.Note, e4.Line, 
       s4.displayName, s4.Note, e5.Line, 
       s5.displayName, s5.Note, e6.Line, 
       s6.displayName, s6.Note, e7.Line, 
       s7.displayName, s7.Note, e8.Line, 
       end.displayName, end.Note

​
Supply Chain — ETO Chip Explorer

Illustrates how to model a manufacturing and supply chain graph. Tracks the flow of materials (e.g. ETO chips) through suppliers, batches, and final assemblies — enabling traceability, root cause analysis, and risk assessment. Query examples: 
# Find all relationships between nodes whose
# descriptions contain "FPGA" using string approximation
MATCH (n{description ~= "FPGA"})-[e]-(m)
RETURN n.displayName, n.description, e, m.displayName, m.description

​
Domain: Healthcare & Life Sciences

​
Reactome Biological Pathways

Explores the Reactome knowledge base of biological pathways. Showcases how TuringDB can rapidly answer multi-hop queries on complex biological interactions such as gene regulation or protein-protein interaction cascades. Query examples: 
# Find all nodes of type "EntityWithAccessionedSequence"
# and referenceType "ReferenceGeneProduct"
MATCH (n:EntityWithAccessionedSequence {"referenceType"="ReferenceGeneProduct"})
RETURN n.displayName, n.schemaClass, n.referenceType

​
Healthcare Knowledge Graph

Builds a patient-centric knowledge graph combining symptoms, diagnoses, treatments, and side effects. Demonstrates entity extraction, GML ingestion, and how TuringDB supports explainable AI on structured health data. Query examples: 
# Find all patients who took Paracetamol
MATCH (p:Patient)-[:TOOK_MEDICATION]->
      (m:Medication {"displayName": "Paracetamol"})
RETURN p.id, p.displayName, m.displayName

​
Running the Notebooks

​
Quick Start

​
Prerequisites

  • Python 3.13 or higher
  • uv package manager

​
Installation

  1. Clone the repository: 
    git clone https://github.com/turing-db/turingdb-examples.git
cd turingdb-examples
  2. Install dependencies: 
    uv sync
  3. Start Jupyter Lab: 
    uv run jupyter lab
    This will start the Jupyter server and display output like: 
    [I 2024-09-26 16:30:15.123 ServerApp] Jupyter Server is running at:
[I 2024-09-26 16:30:15.123 ServerApp] http://localhost:8888/lab?token=abc123...
    Copy the URL (starting with http://localhost:8888/lab?token=...) and paste it into your web browser to access Jupyter Lab.
  4. Open and run the example notebooks in examples/notebooks/public_version/