GOA changes (refresh in commit a3cf70b6d):
- New ISO rows (donor splits of already-reviewed terms): GO:0003948 N4-(beta-N-acetylglucosaminyl)-L-asparaginase activity from pig AGA (RGD:13972209; RGD REST API reports speciesTypeKey 9 = pig) and from human AGA (UniProtKB:P20933); GO:0005764 lysosome from human AGA.
- Retired: GO:0006517 protein deglycosylation (ISO). Human AGA still carries IDA/IMP protein deglycosylation (QuickGO: PMID:1281977, PMID:1904874), so core_functions keeps GO:0006517 as the process.
Actions:
- PENDING resolved: both GO:0003948 ISO donor splits -> ACCEPT (rat enzyme has direct activity data PMID:2775174); lysosome ISO (human donor) -> KEEP_AS_NON_CORE, consistent with the IBA/IEA/MGI-ISO lysosome rows.
- GO:0042802 identical protein binding (IDA, PMID:2775174): action unchanged (MARK_AS_OVER_ANNOTATED) but the rationale was corrected. The earlier reason claimed an alpha2beta2 homotetramer, but the rat abstract says the native enzyme is a heterodimer [PMID:2775174 "The native enzyme had a molecular mass of 49 kDa and was composed of two non-identical subunits joined by strong non-covalent forces"; PMID:1554372 "The native enzyme appeared as a heterodimer among the mammals"]. The alpha and beta chains both come from the P30919 precursor, which probably explains the curator's choice. Full text not cached.
- GO:0008150 biological_process (ND): REMOVE kept; replaced the boilerplate reason with a specific one (the rat enzyme has direct biochemical data, so "no data" is outdated).
- EXP GO:0003948 (PMID:1554372) and IDA GO:0003948 (PMID:2775174): added activity-bearing quotes; actions unchanged (ACCEPT).
- All other rows re-audited; no change. IBA cytoplasm stays KEEP_AS_NON_CORE (lysosome is part_of cytoplasm).
Open questions:
- Does the full text of PMID:2775174 show self-association of whole alpha/beta units in rat (which would support GO:0042802 as stated), or does the annotation just record alpha-beta chain association?
- Why did the rat ISO protein deglycosylation row get withdrawn while the human IDA remains? Possibly an RGD pipeline change. Aga acts on free glycoasparagines (it needs a free alpha-amino and alpha-carboxyl on the Asn), so GO:0006516 glycoprotein catabolic process may fit better than protein deglycosylation.
Review follow-up 2026-10-17 (PR #4617 reviewer findings): GO:0042802 identical protein binding (IDA PMID:2775174) changed from MARK_AS_OVER_ANNOTATED to KEEP_AS_NON_CORE: two identical precursor chains dimerise before autocatalytic cleavage, and UniProt records the mature enzyme as a dimer of alpha/beta heterodimers [UniProtKB:P30919 "SUBUNIT: Heterotetramer of two alpha and two beta chains arranged as a dimer of alpha/beta heterodimers."], so the term is correct for the gene product even though the rat abstract only shows the alpha/beta assembly; this answers the open question above. The ER ISO row reason now explains biosynthetic transit (signal peptide, N-glycosylation) instead of template text. CC convention for this batch: a location row is ACCEPTed only when that location is part of a core_functions entry (Pnlip extracellular region for luminal digestion, Arsb lysosome for GAG degradation); otherwise location rows are KEEP_AS_NON_CORE. Batch policy for broad parents (applied to all eight genes in PR #4617): an uninformative broad parent of a specific activity or process the gene already carries is MODIFIED to that specific term; experimental rows stay at the curator's breadth unless the cited assay itself identifies the specific substrate. (Aga's GO:0016787 MODIFY already followed it.)