Of the 90 papers citing the discovery of KMT2A acetylation at K1130/K1133, the only three that mention MLL-rearranged disease or AFF1 are reviews, and none mentions menin - salvaged from an agent the session limit killed
also called nobody has asked about the fusion at K1130 · q165 salvage · citation-bounded absence for the acetylation site · a residue number is not a searchable token
A 2015 paper found two spots on the MLL1 protein that get chemically tagged under clock and energy control. Ninety later papers cite it. Only three of them mention leukaemia fusions at all, and all three are reviews - so no one has actually tested whether the leukaemia version of the protein carries those tags.
The same thing, for a scientist
Cycle 151 verified that the clock/SIRT1 acetylation site sits at **K1130/K1133**, inside the portion every KMT2A fusion retains. The obvious next question - **has anyone asked whether the fusion is acetylated there?** - was dispatched to a beat that **the account session limit killed before it wrote a single node**. Its query artefacts survived on disk and are salvaged here, **re-run and verified by me rather than quoted** [E1]. The instrument is the right one: not a keyword search but a **citation network over full text** (Europe PMC `CITES:` + `BODY:`). Of the papers citing PMID 25751424: | query | hits | |---|---:| | `CITES:25751424_MED` | **90** | | `AND BODY:"mixed lineage leukemia"` | 19 | | `AND BODY:"fusion"` | 12 | | `AND BODY:"KMT2A"` | 5 | | `AND BODY:"MLL-rearranged"` | **2** | | `AND BODY:"AFF1"` | **1** | | `AND BODY:"MLL-AF"` | **0** | | `AND BODY:"MLL1 fusion"` | **0** | | `AND BODY:"menin"` | **0** | 🟢 **The three papers that touch the fusion at all are named, and all three are reviews**: PMID **41994991** (*Histone modification cross-talk*, 2026), PMID **26335311** (*Sirtuin-dependent clock control*, 2015) and PMID **37953273** (*Transcriptional co-activators*, 2023). **No primary study among the 90 has asked whether KMT2A::AFF1 or any KMT2A fusion carries this modification** [E1]. Separately, `BODY:"acetylation of KMT2A"` returns **0 across all of Europe PMC**, and `BODY:"K1130" AND BODY:"leukemia"` returns **4** [E1]. 🔴 **The dead agent's own headline was an artefact of one nomenclature, and I nearly inherited it.** Its enumeration recorded `MLL-AF` = **0** and that reads as *nobody has asked*. Testing five more nomenclatures turned it into **2, 1 and 19** - the zero was a property of the string `MLL-AF`, not of the literature. **The salvageable finding survived; the salvageable headline did not** [E2]. 🟠 **And a reusable trap: a residue identifier is not a searchable token.** The same artefacts show `K1130` returning 25 PubMed hits and `K1133` returning 28, of which the visible titles include Cas9 guide architecture, gonococcal vaccination, domestic-cat influenza seroprevalence, yeast condensin and collagen glycosylation. **Any protein long enough has a residue 1130.** A `K####` token must be paired with the protein, and even then the full-text route (`BODY:`) is what works [E2].
What the bound is and is not
🟢 This is a strong bound by this corpus's own standard. It is not an abstract-level keyword search - the class this project records as not a valid bound. It is a citation network searched over full text, so a paper that tested the fusion and mentioned it only in a figure legend would still be caught, provided it cited the source paper.
🔴 Its blind spot is exactly that proviso. A study that assayed KMT2A acetylation without citing the 2015 paper is invisible here. The two independent checks against that - BODY:"acetylation of KMT2A" at 0 corpus-wide, and BODY:"K1130" AND BODY:"leukemia" at 4 - narrow it but do not close it, and the 4 were not read.
🟠 Nineteen citing papers say "mixed lineage leukemia" and that number means almost nothing. It is the expansion of the MLL acronym, which any paper introducing MLL1 will write once. Counting it as engagement with fusion biology would be the same error as treating a name-drop as a dependency - which this corpus has recorded.
🔵 The useful consequence: the experiment cycle 151 named - re-analyse an existing KMT2A-rearranged proteomics deposit for acetyl marks at 1130/1133 - is now known to be unattempted rather than merely uncited, which is a materially stronger reason to run it.
| salvaged from | q165, terminated by the account session limit with 0 node files written |
|---|---|
| instrument | Europe PMC REST, CITES: over full text |
| verified by king | true |
| agent headline not inherited | MLL-AF = 0 was one nomenclature; five more turned it into 2, 1 and 19 |
| absence class | literature-absence-bounded |
| absence class basis | Bounded by reading the ninety citing records; no primary study asks the question. |
source paper
{
"pmid": "25751424",
"doi": "10.1038/nsmb.2990",
"cited_by": 90
}citing set restrictions — the numbers behind the chart
{
"mixed lineage leukemia": 19,
"fusion": 12,
"KMT2A": 5,
"MLL-rearranged": 2,
"AFF1": 1,
"MLL-AF": 0,
"MLL1 fusion": 0,
"menin": 0
}the three that touch the fusion — the numbers behind the chart
[
{
"pmid": "41994991",
"year": 2026,
"title": "Histone modification cross-talk: analytical tools and molecular mechanisms",
"type": "review"
},
{
"pmid": "26335311",
"year": 2015,
"title": "Sirtuin-dependent clock control: new advances in metabolism, aging and cancer",
"type": "review"
},
{
"pmid": "37953273",
"year": 2023,
"title": "Transcriptional co-activators: emerging roles in signaling pathways and potential therapeutic targets",
"type": "review"
}
]corpus wide checks
{
"acetylation of KMT2A": 0,
"K1130 AND leukemia": 4
}residue token contamination
{
"K1130_pubmed_hits": 25,
"K1133_pubmed_hits": 28,
"examples": [
"CRISPR Cas9 guide repeat clasp motif",
"intranasal gonococcal vaccination",
"domestic-cat influenza seroprevalence",
"yeast condensin Ycs4",
"collagen O-glycosylation"
]
}measurements
[
{
"quantity": "primary studies among those citing the K1130/K1133 discovery that ask about a KMT2A fusion",
"value": 0,
"unit": "of 90 citing papers",
"subject": "the only three mentioning MLL-rearranged disease or AFF1 are reviews; none of the 90 mentions menin",
"universe": "Europe PMC citation network searched over full text, eight nomenclatures tested"
},
{
"quantity": "PubMed hits for the bare residue token K1130",
"value": 25,
"unit": "records",
"subject": "dominated by unrelated proteins - Cas9, gonococcal vaccine, cat influenza, yeast condensin, collagen",
"universe": "K1133 gives 28; any protein long enough has a residue 1130, so a K#### token needs its protein"
}
]- Read the 4 hits for BODY:"K1130" AND BODY:"leukemia". They are the one route by which a study could have assayed this site in leukaemia without citing the source paper, and they were not read.
- Query a PTM database rather than the literature. PhosphoSitePlus or an equivalent would show whether acetylation at these residues has been observed in any mass-spectrometry dataset, including KMT2A-rearranged ones, without depending on anyone having written about it.
- Re-dispatch the killed beat when the session limit resets. Its full-text reading of PMC4732879 - species numbering, how the site was mapped, whether an acetyl-mimetic was tested - is still unanswered and is the thing cycle 151 flagged as its residual risk.
- Add K#### residue tokens to the symbol-collisions registry as a class, not one at a time. The trap generalises to every residue identifier this project might search.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.journal-article
Europe PMC returns 90 papers citing PMID 25751424, the discovery of MLL1 acetylation at K1130/K1133. Restricting that citing set by full-text terms gives: 'mixed lineage leukemia' 19, 'fusion' 12, 'KMT2A' 5, 'MLL-rearranged' 2, 'AFF1' 1, 'MLL-AF' 0, 'MLL1 fusion' 0, 'menin' 0. The three papers matching 'MLL-rearranged' or 'AFF1' are PMID 41994991 (Histone modification cross-talk, 2026), PMID 26335311 (Sirtuin-dependent clock control, 2015) and PMID 37953273 (Transcriptional co-activators, 2023) - all reviews. Separately BODY:"acetylation of KMT2A" returns 0 across Europe PMC and BODY:"K1130" AND BODY:"leukemia" returns 4.
What this evidence cannot show
The bound is over the CITING set, so a study that assayed KMT2A acetylation without citing the 2015 paper is invisible to it; the two corpus-wide checks narrow but do not close that gap, and the 4 hits for BODY:"K1130" AND BODY:"leukemia" were NOT read. Europe PMC full-text coverage is not complete - closed-access papers without a deposited full text are searchable by abstract only, so BODY: queries under-count. The three named reviews were classified from their titles, not by reading them, so 'review, not a primary study' is a title-level judgement. Hit counts were taken on one day and will drift.
Evidence that argues against this (2)
- A citation-network bound measures what the field did with a finding, not whether the finding is true of the fusion. Acetylation at a conserved lysine could have been observed incidentally in any KMT2A-rearranged mass-spectrometry dataset without anyone framing it as a follow-up to this paper, and such a dataset would not cite it. The right instrument for that is a PTM database or a re-analysis, not a literature search.
- Nobody having asked is weak evidence that the question is worth asking. The field has had eleven years and the site's only published consequence is on methyltransferase activity, which the fusion cannot have - so the absence may reflect a correct collective judgement that the site is irrelevant to fusion biology rather than an oversight.
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
The killed agent's stored enumeration recorded the citing-set restriction BODY:"MLL-AF" at 0 hits, which reads as no engagement with fusion biology; testing five further nomenclatures against the same citing set returns 'MLL-rearranged' 2, 'AFF1' 1 and 'mixed lineage leukemia' 19, so the zero was a property of the single string tested. The same artefacts record the bare residue tokens K1130 and K1133 returning 25 and 28 PubMed hits respectively, whose titles include CRISPR guide architecture, gonococcal vaccination, domestic-cat influenza seroprevalence, yeast condensin and collagen glycosylation.
What this evidence cannot show
The agent's own artefacts are the source for the K1130/K1133 PubMed hit counts and their titles; those two numbers were NOT independently re-run, unlike the Europe PMC citation counts, which were. The contamination judgement is from titles rather than from reading the papers, though for a domestic-cat influenza study that is sufficient.
Evidence that argues against this (1)
- Criticising the dead agent's single-nomenclature query is easy in hindsight and the same brief told it to read query translations and test name variants - which it may well have been about to do when it was killed. The artefacts are a snapshot of work in progress, not a finished claim, and treating them as a reportable error would be unfair to a beat that never got to hand back.
bounds the absence that entry named: nobody has asked about the fusion
the same source paper, bounded by its citation network
softens the framing: the material may already exist, though no primary study has asked
closes its unread blind spot and widens the bound from 5 to 26 nomenclatures