Pax6 (mouse, P63015) curation notes
Provenance / method
- Automated deep research was NOT available for this gene (falcon provider returned HTTP 402, OpenAI 401).
No -deep-research-<provider>.md file was created. These notes were assembled manually from the cached
publications in publications/, the UniProt record (Pax6-uniprot.txt) and the completed human
ortholog review (genes/human/PAX6/PAX6-ai-review.yaml).
- Many cited MGI papers are abstract-only in the cache; for those, the "do not overrule curators" rule was
applied (developmental IMP rows kept as non-core; UNDECIDED where the abstract does not mention Pax6).
Molecular function
Biological roles (mostly non-core, KEEP_AS_NON_CORE)
Problem annotations
- GO:0003682 chromatin binding IDA PMID:23637604: cached full text is a Candida albicans/Dectin-1 paper with no Pax6 mention -> UNDECIDED, reference flagged WRONG_IDENTIFIER.
- GO:0001933 (PMID:16497297, HSP27), GO:0023019 (PMID:18462699, ER71), GO:0007224 (PMID:18590716, FKBP8), GO:0045665 (PMID:16950124),
GO:0042462 (PMID:12648492): abstracts do not support the specific Pax6 claim -> UNDECIDED.
- GO:0007601 visual perception: Pax6 Sey-Dey eyeless mice used only as blind controls -> over-annotation.
- GO:0030216 keratinocyte differentiation -> MODIFY to GO:0160122 corneal epithelial cell differentiation.
- protein binding rows -> REMOVE, except Tle6/Grg6 row -> MODIFY to GO:0001222 transcription corepressor binding.
2026-10-12 Ortholog/family audit (human PAX6, DROME ey/toy, DANRE pax6a/pax6b)
- GOA refresh (
just fetch-gene mouse Pax6): 266 rows fetched; review already covered all 242 distinct (term, evidence, reference) keys; no new or stale rows. No review actions changed in the mouse file.
- PANTHER: mouse and human both PTHR45636:SF48 (PAX-6); fly ey/toy PTHR45636:SF41. Shared family IBAs (brain/forebrain/retina development, sensory organ development, type B pancreatic cell differentiation) are KEEP_AS_NON_CORE in both mouse and human; core MF (GO:0000981, GO:0000978 accepted) agree across mouse, human, ey and toy.
- Mouse-vs-human contradictions resolved on the human side (human IEA rows are Ensembl projections of the mouse experimental rows):
- GO:0003723 RNA binding: mouse IDA KEEP_AS_NON_CORE (Paupar lncRNA association) vs human IEA REMOVE claiming no experimental support. The human projection derives from the mouse IDA PMID:24488179, so the human row was changed to KEEP_AS_NON_CORE.
- GO:0031625 ubiquitin protein ligase binding: mouse IPI KEEP_AS_NON_CORE vs human IEA ACCEPT. PAX6 is the TRIM11 substrate PMID:18628401; human changed to KEEP_AS_NON_CORE.
- Left as-is: GO:0003682 chromatin binding IDA PMID:23637604 stays UNDECIDED (PubMed confirms this PMID is a Candida albicans/dectin-1 paper, so the MGI row likely carries a wrong PMID; the function itself is accepted from other rows). GO:0007601 visual perception is MARK_AS_OVER_ANNOTATED (mouse IMP) vs KEEP_AS_NON_CORE (human TAS); both express "downstream of eye development, not core", so not harmonised.
- Zebrafish pax6a/pax6b ACCEPT several developmental IBAs that mammals keep as non-core, and pax6a REMOVEs the type B pancreatic cell differentiation IBA that pax6b accepts; flagged for the zebrafish reviewers, not edited here.