Deliverome and GO

How GO can describe, and learn from, an atlas of therapeutic delivery addresses

AI Gene Review · projects/DELIVEROME · 2026

Bottom line

  • The Deliverome atlas wants surface proteins that cargo can use as a delivery address: present, internalized, and routed somewhere useful.
  • We defined the term, mapped it to 17 GO anchor terms, and compared human RAB7A with mouse Rab7 as a worked post-uptake routing example.
  • The two reviews agree on core Rab7 biology; mouse adds in vivo evidence that Rab7 loss increases LNP escape. 6 of 9 tasks done.

A delivery route in GO terms

Where GO annotation stops

Deliverome evidence Possible GO use Caution
Surface proteomics, endogenous protein cell surface / plasma membrane needs cell-type context
Endogenous receptor uptake receptor internalization cargo uptake alone is a delivery phenotype
Receptor tracked to lysosome routing annotations not if only the payload was tracked
Perturbation screen hits trafficking machinery review separate direct from viability effects
Expected receptor fails to internalize NOT, rarely failed delivery is not enough

Engineered-cargo outcomes (for example cytosolic escape) belong in GO-CAM-like Deliverome models, not canonical GO.

Worked example: mouse Rab7

Mouse Rab7 review: 119 rows, 80 accepted. Endosome-to-lysosome transport is core, supported by the liver LNP study (PMID:41814093).

Human RAB7A vs mouse Rab7

Human RAB7A Mouse Rab7
Annotations reviewed 124 119
ACCEPT / non-core / REMOVE 70 / 24 / 26 80 / 29 / 8
Best use disease genetics (CMT2B) in vivo delivery routing
  • Shared core: Rab GTPase activity, early-to-late endosome maturation, endosome-to-lysosome transport, retromer binding.
  • Mouse SynGO rows (AMPA-receptor traffic, PMID:24217640) kept non-core: the paper centres on stargazin and AP-2/AP-3A.

Status and next steps

  • ✅ Delivery address defined; GO anchors; model-system stance (human primary, pombe for machinery, mouse for in vivo routing).
  • ✅ RAB7A and Rab7 harmonized; SynGO rows revisited.
  • ⬜ Build a GO-derived human surfaceome prior.
  • ⬜ Pick a small address pilot set (e.g. ASGR1, TFRC, FCGRT).
  • ⬜ Draft a GO-CAM-like delivery-route template.

Read more: projects/DELIVEROME.md · genes/human/RAB7A/ · genes/mouse/Rab7/