Skip to content

QUICKSTART.md

Latest commit

 

History

History
149 lines (113 loc) · 3.96 KB

QUICKSTART.md

File metadata and controls

149 lines (113 loc) · 3.96 KB

Quickstart Guide

This guide gets you from zero to a working graph in 5 minutes using IRIS Community Edition (free, no license required).

Prerequisites

  • Docker (any recent version)
  • Python 3.10+
  • pip install iris-vector-graph

Step 1: Start IRIS

From the repository root:

docker compose up -d

Wait about 30 seconds for IRIS to be ready. You can check:

docker compose ps       # Should show "healthy"

IRIS is now running on:

Default credentials: _SYSTEM / SYS

Step 2: Install the library

pip install iris-vector-graph

Step 3: Initialize the schema

import iris
from iris_vector_graph.engine import IRISGraphEngine

conn = iris.connect("localhost", 1972, "USER", "_SYSTEM", "SYS")
engine = IRISGraphEngine(conn)
engine.initialize_schema()
print("Schema ready.")

This creates all SQL tables and compiles the ObjectScript classes. It's idempotent — safe to run multiple times.

Step 4: Load a demo graph

This demo builds a small knowledge graph of biomedical entities:

nodes = [
    {"id": "gene:BRCA1",      "labels": ["Gene"],    "properties": {"name": "BRCA1",    "type": "tumor_suppressor"}},
    {"id": "gene:TP53",       "labels": ["Gene"],    "properties": {"name": "TP53",     "type": "tumor_suppressor"}},
    {"id": "drug:Olaparib",   "labels": ["Drug"],    "properties": {"name": "Olaparib", "mechanism": "PARP_inhibitor"}},
    {"id": "disease:BRCA",    "labels": ["Disease"], "properties": {"name": "Breast cancer"}},
    {"id": "disease:Ovarian", "labels": ["Disease"], "properties": {"name": "Ovarian cancer"}},
]
engine.bulk_create_nodes(nodes)

edges = [
    {"source_id": "gene:BRCA1",    "predicate": "TARGETS",      "target_id": "drug:Olaparib"},
    {"source_id": "drug:Olaparib", "predicate": "TREATS",        "target_id": "disease:BRCA"},
    {"source_id": "drug:Olaparib", "predicate": "TREATS",        "target_id": "disease:Ovarian"},
    {"source_id": "gene:BRCA1",    "predicate": "ASSOCIATED_WITH","target_id": "disease:BRCA"},
    {"source_id": "gene:TP53",     "predicate": "INTERACTS_WITH", "target_id": "gene:BRCA1"},
]
engine.bulk_create_edges(edges)
print(f"Loaded {len(nodes)} nodes and {len(edges)} edges.")

Step 5: Query the graph

Cypher — find what Olaparib treats:

result = engine.execute_cypher(
    "MATCH (d:Drug {node_id:$id})-[:TREATS]->(dis) RETURN d.name AS drug, dis.name AS disease",
    {"id": "drug:Olaparib"}
)
for row in result.rows:
    print(f"{row[0]} → {row[1]}")

Output:

Olaparib → Breast cancer
Olaparib → Ovarian cancer

Find genes that target a drug:

result = engine.execute_cypher(
    "MATCH (g:Gene)-[:TARGETS]->(d:Drug) RETURN g.name AS gene, d.name AS drug"
)
print(result.rows)

Output:

[['BRCA1', 'Olaparib']]

2-hop path — find diseases reachable from BRCA1:

result = engine.execute_cypher(
    "MATCH (g {node_id:$id})-[*1..2]->(target) RETURN DISTINCT target.name LIMIT 10",
    {"id": "gene:BRCA1"}
)
print([row[0] for row in result.rows])

Output:

['Olaparib', 'Breast cancer', 'Ovarian cancer']

Step 6: Explore further

Feature Method Guide
Vector search engine.ivf_build(), engine.search_nodes_by_vector() Python SDK
Temporal edges engine.create_edge_temporal(), engine.get_edges_in_window() Architecture
Graph analytics engine.run_pagerank(), engine.run_khop() Python SDK
REST / Bolt API uvicorn api.main:app Operations
IRIS Management Portal http://localhost:52773 (browser)

Stopping IRIS

docker compose down        # Stop container, keep data
docker compose down -v     # Stop and wipe all data