Skip to content

Latest commit

 

History

History
88 lines (60 loc) · 3.84 KB

File metadata and controls

88 lines (60 loc) · 3.84 KB

Data provenance

Dataset

Field Value
Data version opentargets-24.06-frozen-slice-v2+gapforge-flurizan+astegolimab
Release date 2024-06-01
Scope Frozen slice: 500 genes, ~3,000 disease–target associations with decomposed evidence; plus GapForge educational case studies (Flurizan AD, Astegolimab COPD)
Inspiration Open Targets Platform association + evidence model

GapForge case studies

Field Value
Seed files data/gapforge/*.json (Flurizan + Astegolimab)
Loader scripts/seed_gapforge.py (also invoked at end of seed_neo4j.py)
Framing Historical educational hypotheses — not clinical advice
Public refs ClinicalTrials.gov; PMC / public trial-design discussion

Bulk ingest source (production path)

Field Value
Provider Open Targets Platform (EMBL-EBI & Wellcome Sanger Institute)
Release pinned 24.06 (June 2024)
FTP base https://ftp.ebi.ac.uk/pub/databases/opentargets/platform/24.06/output/etl/json/
Association file associationByOverallDirect — target–disease scores
Evidence file evidence — datatype + datasource rows per association
Documentation Open Targets datasets
Licence CC0 1.0 for platform data (terms); cite Open Targets in publications

Download and load (full ingest)

py -3 scripts/download_opentargets_bulk.py --release 24.06 --max-genes 500
py -3 scripts/etl_opentargets.py --input data/raw/opentargets_bulk.json
py -3 scripts/seed_neo4j.py

Frozen slice (CI / offline demo)

CI and default Docker seed use a deterministic frozen subset derived from the same schema (not a live FTP pull):

py -3 scripts/build_frozen_slice.py   # writes data/raw/opentargets_slice_v2.json + processed CSV
py -3 scripts/seed_neo4j.py

Fixture copies for tests: api/tests/fixtures/opentargets_slice_v2.json, api/tests/fixtures/gene_catalog.json.

Sources

  • Open Targets Platform 24.06 — bulk JSON/Parquet releases via EBI FTP
    Licence: CC0 1.0; see platform-docs.opentargets.org/licence
  • Frozen slice v2 — reproducible subset for CI (see scripts/build_frozen_slice.py)

Scientific scope

  • Graph edges are disease–target associations with an overall score (0–1) plus decomposed evidence (evidence_type, source, optional study_id).
  • Scores indicate correlative strength in curated public data, not causation, mechanism, or clinical actionability.
  • This system does not provide diagnosis, treatment recommendations, or regulatory-grade evidence.

Schema

Relationship properties on ASSOCIATED_WITH: score, source, evidence_type, evidence_json, optional study_id.
See docs/SCHEMA_MIGRATION.md.

API

Live metadata: GET /api/v1/meta (also exposed as MCP resource bioinsight://meta when using embabel-mcp).

Evidence breakdown: GET /api/v1/genes/{id}/evidence (consumed by embabel-mcp get_target_evidence).

Citations

If you reference this graph in a write-up, cite Open Targets and note the release/slice:

Open Targets Platform, EMBL-EBI & Wellcome Sanger Institute. Release 24.06 (or frozen slice v2 for demo/CI).

Updates

When ingesting a new Open Targets release:

  1. Update scripts/download_opentargets_bulk.py default --release
  2. Update api/app/metadata.py (DATA_VERSION, RELEASE_DATE, SOURCES)
  3. Update this file and re-run ETL + scripts/seed_neo4j.py