AFF1/AFF4-CAMKV - DECLINED, and it is a LABELLED REDISCOVERY: data/symbol-collisions.json already holds a CAMKV row saying BioPlex 2017 and 2021 are one programme and that CAMKV is deposited under split isoform accessions
also called AFF1-CAMKV · AFF4-CAMKV · CaM kinase-like vesicle-associated protein · AFF-CAMKV
A brain vesicle protein shows up with both AFF scaffolds, in three separate large surveys, in two different cell lines. It is the most repeated hit on the list and it is still the same kind of experiment done three times.
The same thing, for a scientist
**Verdict: DECLINED, and this is a LABELLED REDISCOVERY, not a new find.** `data/symbol-collisions.json` already carries a CAMKV row, added 2026-09-22 by the King from Q104's labelled rediscovery of a cycle-54 finding: *BioPlex isoform splitting is SYSTEMATIC, not incidental -- ARRB2 and CAMKV in the same publication pair* [E4]. It records that CAMKV is deposited under **an isoform accession in BioPlex 2017 and the canonical accession in BioPlex 2021**, that any accession-keyed BioPlex count in this project is therefore a **floor**, and that **BioPlex 2017 + 2021 are ONE programme, not two independent labs**. **I reached the second of those conclusions independently and should have read the registry first; rule 88.** The row itself: IntAct holds **five** AFF1/AFF4-CAMKV records across **three publications** and **two cell lines** [E1] - BioPlex 2.0 (PMID 28514442, HEK293T, AFF1 and AFF4), BioPlex 3.0 (PMID 33961781, HEK293T, AFF1 and AFF4, AFF4 at miscore 0.53), and the Schaffer U2OS multimodal map (PMID 40205054, AFF4, 0.53). **Two baits, three publications, two lines - no other uncovered prey-only partner has that, and it is still effectively two platforms and one method.** Declined on that, and on compartment: UniProt Q8NCB2 (501 aa) records **Cell membrane** and **Cytoplasmic vesicle membrane**, both only **ECO:0000250 by similarity** - **no experimental localization at all** [E2] - for a CaM-kinase-like pseudokinase characterised in neurons.
READ THE REGISTRY FIRST - what was already written down about this exact protein
data/symbol-collisions.json has 213 rows and one of them is about CAMKV. It was added 2026-09-22, two cycles before this beat, and it says:
ARRB2: deposited as P32121-3 (isoform 3) in BioPlex 2017 and canonical P32121 in 2021. CAMKV: the identical split, found in cycle 54, in the SAME publication pair. Consequence: an accession-keyed filter sees one of the two rows, so every accession-keyed BioPlex count in this project is a FLOOR.
Prescription: Normalise accessions to canonical (strip -N isoform suffixes) before deduplicating or counting BioPlex rows, and treat BioPlex 2017 + 2021 as one programme when assessing lab independence.
Two consequences, one good and one chastening.
Good: this beat's pull already implements the first half of that prescription without knowing it existed. analyses/q160-the-prey-only-interactome/intact-rest.py normalises every partner identifier with theirs.split('-')[0] before grading, so isoform-split rows collapse to one accession and the five CAMKV records here are five records of one protein, not an artefact of the split. The census is not subject to the floor the registry warns about.
Chastening: the second half of the prescription - treat BioPlex 2017 + 2021 as one programme - is the entire argument of the section below, which I worked out from scratch. It was already written down, about this protein, with a measurement attached. Rule 88's corollary says a rediscovery is still useful because it shows the trap is live and catches beats - it caught this one - but it is a different claim from a new find and the entry has to say which. This one is a rediscovery.
And ARRB2 is also in this census's uncovered tail (P32121, AFF4, PMID 33961781), which means the registry's paired example is sitting unwritten four rows below CAMKV in the same ranking.
Reproduction is not independence, and this row is the clean demonstration
This corpus keeps arriving at the same point by different routes:
interface-menin-cd226-declined-proteome-scale-ap-ms: two platforms agreeing on an over-expressed surface receptor in three non-expressing lines is agreement about a shared artefact hypothesis.- Rule 92's note that most AFF4 entries in the read queue trace to just two papers, so a '2 publications' grade overstates independence.
- The four AFF4 two-hybrid partners that are one S. cerevisiae screen split across three PSI-MI labels.
CAMKV is the strongest version of the pattern, because here the reproduction is genuinely across two baits. AFF1 and AFF4 are paralogous scaffolds; recovering the same prey on both is what a real complex member would do - and it is also exactly what a shared background contaminant of AFF-family pull-downs would do, because the two baits share sequence, share purification behaviour and were tagged the same way in the same laboratory.
The discriminator is not more AP-MS. It is a different method or a different lysate: a reciprocal CAMKV-bait pull-down recovering AFF1, or an endogenous co-IP, or any experiment in a cell that expresses CAMKV natively. None exists.
A partner with no experimental localization anywhere
CAMKV is the only partner this beat looked at whose UniProt subcellular-location block is entirely ECO:0000250, by similarity - inferred from an orthologue, with no human experiment behind either line 2. There is no tissue-specificity comment either.
Rule 91 clause 4 is about distinguishing experimental ECO:0000269 from curator-inference ECO:0000305. This is a third tier below both: inference from another species. A localization field populated from this record would carry cell membrane and cytoplasmic vesicle membrane for a protein whose human localization has, as far as UniProt records, never been measured.
And that cuts both ways, which is why the decline is argued rather than asserted. An unmeasured localization cannot supply a compartment-mismatch argument. The honest statement is not CAMKV is in the wrong compartment; it is nobody knows where human CAMKV is, so the compartment check that killed CD226 is unavailable here and the decline rests on platform non-independence instead.
| partner a | gene-aff1 |
|---|---|
| partner b | CAMKV |
| partner b uniprot | Q8NCB2 |
| partner b length AA | 501 |
| partner a uniprot | P51825 |
| partner a length AA | 1210 |
| also recovered with | gene-aff4 (Q9UHB7), three of the five records |
| verdict | DECLINED |
| adjudicated by | q160, cycle 146, 2026-09-24 |
| druggability | Not assessable. CAMKV is a pseudokinase with a degenerate catalytic site and no chemical probe in this corpus. |
| n methods | 1 |
| n platforms | 2 |
| n publications | 3 |
| n records | 5 |
| n baits | 2 |
| expansion method | spoke expansion on the BioPlex and Schaffer rows as this corpus records for the same publications elsewhere; not flagged on these particular rows |
| endogenous side | no - HA/FLAG tandem-tagged baits throughout |
| intact miscore | 0.53 |
| cell state | unstimulated, asynchronous |
| epigenetics | None. |
| decline reason class | maximal reproduction within one method and effectively two platforms; two paralogous baits share artefact as well as biology |
| absence class | never-measured |
| absence class note | No reciprocal CAMKV-bait experiment, no endogenous co-IP, and no experiment in any cell that natively expresses CAMKV. A single reciprocal pull-down would be decisive and none exists. |
| registry row | data/symbol-collisions.json: 'BioPlex isoform splitting is SYSTEMATIC, not incidental -- ARRB2 and CAMKV in the same publication pair', added 2026-09-22. This entry is a labelled rediscovery of its independence argument. |
| isoform normalisation applied | true |
interface residues a
[]
structures
[]
ligands
[]
records
[ "AFF1-CAMKV, anti tag coip, HEK293T, 0.35, PMID 28514442", "AFF1-CAMKV, anti tag coip, HEK293T, 0.35, PMID 33961781", "AFF4-CAMKV, anti tag coip, HEK293T, 0.35, PMID 28514442", "AFF4-CAMKV, anti tag coip, HEK293T, 0.53, PMID 33961781", "AFF4-CAMKV, anti tag coip, U2OS, 0.53, PMID 40205054" ]
cell systems
[ "HEK293T embryonic kidney", "U2OS osteosarcoma" ]
localization
{
"compartment": "cell membrane and cytoplasmic vesicle membrane, BOTH ECO:0000250 by similarity only",
"subnuclear": [],
"experimental_localization_exists": false,
"shown_for_fusion": "NO - nobody has looked in any fusion, and no human experimental localization for CAMKV exists in UniProt at all."
}measurements
[
{
"quantity": "count",
"subject": "IntAct records for CAMKV against the six core baits",
"value": 5,
"units": "records",
"n": 135,
"universe": "the 135 uncovered human prey-only partners, IntAct REST 2026-09-24",
"null": null,
"note": "The highest record count among the uncovered tail. All five are anti-tag co-IP."
},
{
"quantity": "count",
"subject": "distinct AP-MS PLATFORMS behind those five records",
"value": 2,
"units": "platforms",
"n": 3,
"universe": "PMIDs 28514442 (BioPlex 2.0), 33961781 (BioPlex 3.0), 40205054 (Schaffer U2OS map)",
"null": null,
"note": "Three publications, two platforms - BioPlex 2.0 and 3.0 are one project at two releases. The publication count overstates independence by 50%."
}
]- Declining the most-reproduced row in the tail sets a bar that could be used to decline almost anything. The bar being applied is specifically 'one method across correlated platforms', and it would NOT decline a row with a reciprocal pull-down or an orthogonal method - that distinction has to survive, or the criterion becomes unfalsifiable.
- ARRB2 (P32121), the registry's paired example, is also in this census's uncovered tail as an AFF4 prey row from BioPlex 3.0 and has no entry. Writing it would complete the pair the registry names.
- Does a reciprocal CAMKV-bait pull-down recover AFF1 or AFF4? Five records and not one of them runs in that direction.
- Where is human CAMKV actually located? UniProt has no experimental localization at all, which makes it one of the least characterised proteins this corpus has touched.
- How many other rows in this graph rest on BioPlex 2.0 plus BioPlex 3.0 and are counted as two publications? The pattern has now been identified three times in three separate beats and never counted corpus-wide.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
IntAct REST returns five CAMKV (Q8NCB2) records against the six core baits: AFF1 prey / anti-tag co-IP / HEK293T / miscore 0.35 / PMID 28514442; AFF1 prey / anti-tag co-IP / HEK293T / 0.35 / PMID 33961781; AFF4 prey / anti-tag co-IP / HEK293T / 0.35 / PMID 28514442; AFF4 prey / anti-tag co-IP / HEK293T / 0.53 / PMID 33961781; AFF4 prey / anti-tag co-IP / U2OS / 0.53 / PMID 40205054. All five are anti-tag co-immunoprecipitation typed 'association'. PMIDs 28514442 and 33961781 are BioPlex 2.0 and BioPlex 3.0, the same platform and laboratory at two releases.
IntAct molecular interaction database, REST findInteractions on P51825, retrieved 2026-09-24 by q160 (partner CAMKV) · IntAct (EMBL-EBI) 2026What this evidence cannot show
Five records but one method. Three publications but two platforms, and two of the three are releases of one project. Two cell lines, neither of which expresses CAMKV natively as far as any record shows. The row count is the highest in the uncovered tail and is the least informative thing about it.
Evidence that argues against this (1)
- Recovery on TWO paralogous baits is not nothing. If CAMKV bound only AFF4 it would be easier to dismiss; recovering it on AFF1 as well is consistent with a genuine contact at a conserved surface. The counter to that counter is that shared sequence also means shared artefact, and no experiment distinguishes them. Huttlin EL et al. 2021
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
UniProtKB Q8NCB2 (CAMKV, CaM kinase-like vesicle-associated protein) is 501 aa, reviewed. Its subcellular location block records Cell membrane and Cytoplasmic vesicle membrane, BOTH with evidence code ECO:0000250 (by similarity) and NEITHER with any experimental evidence code. No tissue-specificity comment is recorded.
UniProtKB entry Q8NCB2, retrieved 2026-09-24 · UniProt Consortium 2026What this evidence cannot show
The absence of ECO:0000269 is the claim and is declared as such. It means UniProt curators found no human experimental localization to cite, not that one does not exist in the literature - a PubMed search for CAMKV localization was not run here.
Evidence that argues against this (1)
- An absent experimental localization removes the compartment-mismatch argument entirely. A reader could reasonably say that declining CAMKV on compartment grounds is declining it on an orthologue's behalf, which is why this entry declines it on platform non-independence instead and says so. UniProt Consortium 2026
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
PMID 33961781 (BioPlex 3.0) reports AP-MS of 10,128 human proteins in 293T cells giving 118,162 interactions among 14,586 proteins, plus a second network from 5,522 immunoprecipitations in HCT116, and its stated headline is that interactions are extensively cell-line-specific and that subnetworks rewire between the two lines. PMID 28514442 (BioPlex 2.0) is the preceding release of the same project.
Dual proteome-scale networks reveal cell-specific remodeling of the human interactome. · Huttlin EL, Bruckner RJ, Navarrete-Perea J, Cannon JR, Baltier K, Gebreab F, Gygi MP, Thornock A, et al. (28 authors), Gygi SP · Cell 2021What this evidence cannot show
Abstract read in full; interaction tables not opened. This corpus already cites this PMID repeatedly, including in interface-menin-cd226-declined-proteome-scale-ap-ms where the same non-independence argument is made about the same platform - so the reasoning here is applied, not new.
Evidence that argues against this (1)
- BioPlex recovers an enormous number of real complexes and its scale is a strength, not a defect. A row recovered in BOTH BioPlex releases is more likely to be reproducible within the platform, which is a real if limited form of evidence. Huttlin EL et al. 2017
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
data/symbol-collisions.json, added 2026-09-22 by king-cycle-57 from a labelled rediscovery by Q104 of a cycle-54 finding, carries the row 'BioPlex isoform splitting is SYSTEMATIC, not incidental -- ARRB2 and CAMKV in the same publication pair'. It records that ARRB2 was deposited as P32121-3 in BioPlex 2017 and canonical P32121 in 2021, that CAMKV shows the identical split in the same publication pair, that every accession-keyed BioPlex count in this project is therefore a floor, and that BioPlex 2017 and 2021 are one programme rather than two independent laboratories. Its prescription is to normalise accessions to canonical before counting and to treat the two releases as one programme when assessing independence.
What this evidence cannot show
This is a project-internal registry, not a primary source; the underlying BioPlex deposition accessions were not re-checked live here. The registry itself labels the CAMKV finding as a cycle-54 discovery rediscovered by Q104 in cycle 57, so it has already survived one independent rediscovery before this one.
Evidence that argues against this (1)
- The registry's headline concern - accession-keyed counts being a floor - does NOT apply to this census, because intact-rest.py strips isoform suffixes before grading. Quoting the row as a caution against this beat's numbers would be wrong; it is quoted as the prior art for the independence argument, which is a different part of the same row. this project 2026
prey in AFF1 pull-downs, BioPlex 2.0 and 3.0, HEK293T; DECLINED
prey in AFF4 pull-downs across BioPlex 2.0, 3.0 and the Schaffer U2OS map; DECLINED
the most-reproduced row among the 135 uncovered prey-only partners, and still one method