Virtual Embryo — atlas data + knowledge graph

Virtual Embryo — atlas data + knowledge graph

Virtual Embryo is an AI-augmented, multi-modal atlas of mouse (and human) embryonic development built in the Xiaojie Qiu lab. It unifies, under one anatomical reference frame: a knowledge graph (genes · anatomy · developmental stages · expression · disease · drugs · papers, ~1.8M facts in Neo4j) and a 3D data atlas (eMouseAtlas OPT/histology reference volumes + anatomy meshes, and 3D spatial-transcriptomics reconstructions).

This skill lets you answer developmental-biology questions from the KG and pull atlas datasets and render them in 3D with the volume3d / spatial3d desktop viewer apps.

The public interfaces (read-only, no key)

All reads are open (no auth). The four write endpoints (/kg/cypher_write, /kg/submit_extraction, …) need an admin key and are not for general use. Quick health check: curl https://kg.virtualembryo.ai/healthz → {"ok":true}.

1. Knowledge graph — query the developmental KG

Hit the endpoints with plain HTTP (requests). All return JSON.

import requests
KG = "https://kg.virtualembryo.ai/kg"

# Resolve a name → canonical entity (gene / anatomy / stage / disease …)
hits = requests.get(f"{KG}/search", params={"q": "Sox2", "limit": 5}).json()
# → {"results":[{"iri":"http://identifiers.org/mgi/MGI:98364","label":"Sox2",
#                "type":".../Gene","match_kind":"exact"}]}
iri = hits["results"][0]["iri"]

# One-hop details (type, synonyms, direct relations)
requests.get(f"{KG}/entity", params={"iri": iri}).json()

# Where is a gene expressed? (curated anatomy × stage, from paper extractions)
requests.get(f"{KG}/expression", params={"gene": "Sox2", "stage": "TS17"}).json()
# → {"expressions":[{"anatomy_label":"neural tube","stage":"TS17",
#                    "intensity":...,"paper_doi":...,"confidence":...}], ...}

# Neighbours of an entity, optionally one relation type
requests.get(f"{KG}/expand", params={"iri": iri, "rel": "EXPRESSED_IN"}).json()

# A subgraph for reasoning / a graph view (anchor on a stage or seed entity)
requests.get(f"{KG}/subgraph", params={"stage": "TS17", "limit": 200}).json()
requests.get(f"{KG}/subgraph", params={"seed": iri, "limit": 100}).json()

# Read-only Cypher escape hatch (write keywords are blocked server-side)
requests.post(f"{KG}/cypher", json={
    "cypher": "MATCH (g:Gene)-[:EXPRESSED_IN]->(a:Anatomy {name:'neural tube'}) "
              "RETURN DISTINCT g.name LIMIT 50"}).json()