ADPRS (Q9NX46) - computed evidence for the GO annotation review

Generated by verify_adprs_claims.py --render from results.json. Do not hand-edit:
--check fails if this file and results.json disagree.

Data fetched: 2026-07-27

1. PANTHER node reach (fully paginated)

node	annotations	gene products	terms	every member same terms	includes Q9NX46	reaches ADPRH/ADPRHL1
`PANTHER:PTN001045209`	128	64	GO:0005634, GO:0005739	True	True	no
`PANTHER:PTN008564042`	132	33	GO:0004649, GO:0071451, GO:0140290, GO:0140292	True	True	no

The last column is tested on accessions (P54922 human ADPRH, Q8NDY3 human ADPRHL1),
not on symbol spellings. Neither node reaches either paralogue - which is the result
that matters for a family whose three members have different specificities (ADPRH:
arginine; ADPRS: serine/PAR/O-acetyl-ADP-ribose; ADPRHL1: catalytically inactive).
The remaining members are ADPRS orthologues; those without an ADPRS/adprs/ADPRHL2
symbol are unnamed loci in non-model genomes:

PTN001045209 other symbols (37): 1273432, A0A1I8FBP4, A0A5S6QJY6, A0A7M5UKJ7, A0A7M5VG22, A0A7M7NW39, A0AA85FH88, ACJMK2_005712, AFUS01_LOCUS18876, AUGUSTUS-3.0.2_01093, Adprhl2, BpHYR1_020224, CDAUBV1_LOCUS12368, CDAUBV1_LOCUS12370, DAPPUDRAFT_306696, DGYR_LOCUS2331, EmuJ_001181500, Fcan01_02547, H2XRC4, HPB48_002562, LARSCL_LOCUS10237, LOC100199172, LOC118403597, NEMVEDRAFT_v1g229067, ODALV1_LOCUS6645, OXX778_LOCUS2713, PACLA_8A085503, RvY_03423-1, SBAD_LOCUS8361, TCAL_10085, TRIADDRAFT_24625, TRIADDRAFT_55486, TRIADDRAFT_63844, TTAC_LOCUS4738, Vbra_13646, W5LXM9, WBGene00111238
PTN008564042 other symbols (13): 1273432, A0A7M7NW39, ANIA_04010, AUGUSTUS-3.0.2_01093, DAPPUDRAFT_306696, H2XRC4, LOC118403597, NEMVEDRAFT_v1g229067, TRIADDRAFT_24625, TRIADDRAFT_55486, TRIADDRAFT_63844, W5LXM9, WBGene00111238

2. The superoxide annotation names the wrong reactive oxygen species

GO:0070301 a descendant of GO:0071451: False
GO:0071451 a descendant of GO:0070301: False
share GO:0034614 cellular response to reactive oxygen species: True

So they are siblings; the current annotation is not a safe generalisation of the data.

The donor of the IBA is mouse Adprs (Q8CG72, 56 annotations). Its
GO:0071451 IMP cites PMID:24191052.

cached paper	contains 'hydrogen peroxide'	contains 'superoxide'
PMID:24191052	True	False
PMID:30401461	True	False

Both experiments behind the term used hydrogen peroxide; neither mentions superoxide.

3. The cell-death term: parthanatos is not necroptosis

mouse GO:0060546 IDA cites PMID:30830864
GO:0062099 is an ancestor of GO:0060546: True (so generalising asserts nothing new)
GO:0070266 definition invokes RIPK1/3: True; MLKL: True
PMID:34479984 cached text contains 'induction of cell death via the parthanatos pathway': True

GO records the ambiguity itself, in the comment on GO:0097527:

This is sometimes referred to as PARP-dependent cell death or parthanatos; it is still being debated if it constitutes an independent cell death modality.

4. A mitochondrial reaction inheriting a nucleoplasmic BER term

ADPRS's only Reactome reaction: R-HSA-8952903 = ADPRHL2 hydrolyses poly(ADP-ribose)
its compartment: mitochondrial matrix
its inputs: poly(ADP-ribose) [mitochondrial matrix], H2O [mitochondrial matrix]
containment chain: R-HSA-8952903 > R-HSA-110362 > R-HSA-110373 > R-HSA-73933 > R-HSA-73884 > R-HSA-73894

parent pathway	name	compartment	goBiologicalProcess
`R-HSA-110362`	POLB-Dependent Long Patch Base Excision Repair	nucleoplasm	None
`R-HSA-110373`	Resolution of AP sites via the multiple-nucleotide patch replacement pathway	nucleoplasm	0006287

Reference-projection test on Reactome:R-HSA-110373: 5 annotations over 5 entities - ADPRS, FEN1, LIG1, PARG, POLB - all to GO:0006287.

Contrast the gene-specific reaction Reactome:R-HSA-8952903: 2 annotations over 1 entity. A single-gene curated reaction, not a bulk import.

5. Where GO places the ADP-ribosyl hydrolase activities

term	label	acceptor atom	under GO:0016799 (N-glycosyl)	under GO:0004553 (O-glycosyl)
`GO:0003875`	ADP-ribosylarginine-[protein] hydrolase activity	arginine guanidino N (N-glycosidic)	True	False
`GO:0004649`	poly(ADP-ribose) glycohydrolase activity	ribose-ribose 1''-2' O (O-glycosidic)	False	True
`GO:0140292`	ADP-ribosylserine-[protein] hydrolase activity	serine hydroxyl O (O-glycosidic)	True	False
`GO:0140293`	ADP-ribosylglutamate-[protein] hydrolase activity	glutamate carboxylate O (O-glycosidic)	True	False

GO:0003875 is the internal positive control: arginine's acceptor really is a nitrogen, so
its N-glycosyl placement is right. The serine and glutamate terms share that placement while
their acceptor atoms are oxygens. Reported as a question for GO, not acted on here.

6. Both protein-binding rows are one two-hybrid screen

partner	IntAct records from PMID:32296183	detection methods
PRDM5	3	two hybrid array; two hybrid prey pooling approach; validated two hybrid
TNKS	3	two hybrid array; two hybrid prey pooling approach; validated two hybrid

accession	entry	reviewed	length	IntAct partners
O95271	TNKS1_HUMAN	True	1327	186
Q9NQX1	PRDM5_HUMAN	True	630	96
Q9NX46	ADPRS_HUMAN	True	363	35

The three records per partner are sub-methods of one HuRI screen, which is what UniProt's
NbExp=3 counts. Both partners resolve to reviewed canonical entries, so neither is a
TrEMBL/ORFeome substitution - a check reported as negative rather than omitted.

7. Retraction / erratum scan

Scanned 26 PMIDs; 26 resolved; unresolved: none.

PMID	type	correction
30045870	ErratumIn	PMID:30659162 - J Biol Chem. 2019 Jan 18;294(3):874. doi: 10.1074/jbc.AAC118.007258.
30100084	ErratumIn	PMID:30388405 - Am J Hum Genet. 2018 Nov 1;103(5):826. doi: 10.1016/j.ajhg.2018.10.002.
30100084	ErratumIn	PMID:34861176 - Am J Hum Genet. 2021 Dec 2;108(12):2385. doi: 10.1016/j.ajhg.2021.11.013.

No retractions. All three errata were retrieved and read: PMID:30659162 corrects one
author's affiliation on PMID:30045870, and PMID:30388405 / PMID:34861176 are author-list
corrections to PMID:30100084. None touches data.

What this script does NOT establish

The claim that PMID:33769608 contains no poly(ADP-ribose) experiment is a reading of the
full text, not a computation. What is mechanised is only that the sentence the review
quotes is present verbatim in the cached copy (True). A phrase-presence check cannot prove the
absence of an experiment, and saying so is more useful than a check that looks like it can.