The extracellular matrix

GO annotation review of ten human basement-membrane and interstitial matrix genes

AI Gene Review · projects/ECM · IN_PROGRESS · 2026

Bottom line

  • All 10 planned genes are reviewed across three phases (the task list on the page still shows Phase 3 open).
  • 639 GOA rows: 284 ACCEPT, 193 non-core, 73 REMOVE, 63 MODIFY, 17 NEW.
  • FN1 carries most corrections: 52 protein binding rows removed, 38 extracellular region rows sharpened to extracellular matrix.

Where the genes sit

Why the ECM

  • ECM proteins are large, multi-domain and heavily annotated from interaction screens, so generic rows pile up.
  • Their core role is structural and extracellular, while many rows describe downstream developmental outcomes.
  • The testicans (SPOCK1-3) test whether a review can capture a paralog with the opposite effect.

Results by gene

What a review looks like

FN1 review page: nervous system development is kept as non-core; cell-substrate junction assembly is accepted as core.

Key findings

  1. FN1: 52 IPI protein binding rows REMOVE; 38 extracellular region rows MODIFY → extracellular matrix.
  2. SPOCK2: metalloendopeptidase inhibitor rows MODIFY (→ endopeptidase regulator / negative regulation of endopeptidase activity), because SPOCK2 counter-inhibits SPOCK1/3.
  3. EPYC: NEW collagen binding, collagen fibril organization, extracellular matrix structural constituent; TAS glycosaminoglycan binding REMOVE.
  4. SPOCK3: NEW collagen binding, extracellular matrix binding, negative regulation of endopeptidase activity.

Status and next steps

  • ✅ 10/10 planned gene reviews complete.
  • ⬜ DCN has no core_functions block yet.
  • ⬜ No ECM module; basement membrane (NID1, HSPG2, AGRN with laminin and collagen IV) is the obvious first one.
  • ⬜ Update the stale task list on the project page.

Read more: projects/ECM.md · genes/human/FN1/ · genes/human/SPOCK2/