C1GALT1 (T-synthase) — curation notes

2026-09-17 — de novo review

No -deep-research-PROVIDER.md: tooling unavailable (OpenAI key rejected);
nothing self-authored was named as provider output.

The committed enzyme was missing from its own pathway

C1GALT1's only two biological-process rows in GOA were angiogenesis and
kidney development — both cited to a case-control SNP association study. It had
no mucin-type O-glycosylation process annotation at all, despite performing
the committed galactosylation step.

Ran the comparator check before proposing anything, per CLAUDE.md:

• GALNT1 (immediately upstream) carries GO:0016266 ✓
• GCNT1 (immediately downstream) carries GO:0016266 ✓
• C1GALT1C1 / Cosmc — its own non-catalytic chaperone — carries the parent
  GO:0006493 three times ✓

A pathway whose initiating enzyme, branching enzyme and a chaperone are all
annotated to the process, but whose committed enzyme is not, is an omission
rather than a curatorial line. GO:0016266 added as the single NEW. The
participation test is met directly — C1GALT1 catalyses a step — and the term's
definition explicitly covers elongation ("which can be further elongated with the
sequential addition of").

Citation problem the format checks cannot see

PMID:17228361 is correctly identified, correctly titled, and supports neither
annotation drawn from it. It is a case-control association study of 670 IgAN
patients and 494 controls carrying an IMP evidence code. Recorded
correctness: MISCITED.

Angiogenesis vs kidney development — why they differ

Both rows cite the same unsupporting reference, so the asymmetry needs its own
evidence. It has it on one side only:

• Angiogenesis → KEEP_AS_NON_CORE. The T-synthase knockout has a defined
  developmental vascular phenotype: PMID:14745002 and PMID:14745002 Real but downstream —
  the consequence of altered glycans on other proteins, not something this enzyme
  does during angiogenesis.
• Kidney development → MARK_AS_OVER_ANNOTATED. No supporting phenotype at
  all; IgA nephropathy is an acquired adult glomerular disease driven by
  under-galactosylated IgA1 from plasma cells, not a developmental defect.

(An earlier draft justified the angiogenesis call by an uncited appeal to "the
mouse literature". That was flagged in PR review on #3058 and is why
PMID:14745002 was fetched and cited. Recording it here: an unreferenced appeal to
a body of literature is not support, even when the literature exists.)

Client of a chaperone

The protein binding row (partner Q96EU7 = Cosmc) was MODIFYed to
GO:0051087 protein-folding chaperone binding rather than removed — unusually
for a generic protein binding row, the cited paper supports a specific
client-side MF: PMID:21496458 The reciprocal chaperone
activity sits on Cosmc.

Compartment

Three GO:0016020 membrane rows collapsed to GO:0000139 Golgi membrane.
PMID:11677243 gives the topology; PMID:37216524 gives the organelle.
Note the contrast with its chaperone, which is ER-resident — the compartment
split between the two is explicit in the same sentence pair.

Cross-reference

HPA calls C1GALT1 "Nuclear bodies, Cytosol" with no Golgi call — the same
replace-the-Golgi pattern that made GCNT1's nuclear speck annotation an artefact.
Unlike GCNT1's, this one was never imported into GO. See GCNT1-notes.md.