Oncofusion

start / graph / ana-the-infant-menin-inhibitor-occupancy-experiment-exists-and-it-is-kmt2a-enl-not-aff1

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.

The infant menin-inhibitor occupancy experiment has been done and deposited - menin, KMT2A and ENL CUT&RUN under revumenib in ONE infant patient sampled before and after a lineage switch, not two patients - but it is KMT2A-ENL, so the index-fusion absence survives, and the same deposit holds untreated KMT2A::AFF1 libraries this corpus cites once

also called GSE252378 has infant menin-inhibitor menin CUT&RUN · the infant experiment exists in the wrong fusion · Janssens 2024 Nat Commun holds the arm this project said was missing · 25 KMT2A-AFF1 primary CUT&RUN libraries cited by one entry

This project has repeatedly recorded that nobody has measured where menin sits on the DNA of an infant's leukaemia cells while a menin-blocking drug is present. That experiment turns out to have been done and made public — in two infant patients — but their leukaemias carry a different version of the same broken gene than the one this project studies. The gap is real, and it is now much narrower and much better described.

The same thing, for a scientist

🔴 **CORRECTED AGAIN 2026-09-24 (cycle 128): ONE ALIQUOT, NOT TWO, IS UNDER THE DRUG — AND THE OTHER IS ITS OWN PRE-TREATMENT BASELINE.** GEO labels **both** aliquots *"treated with the Menin inhibitor"*, but that is a **case-level** string. The paper is explicit, and I read it directly (PMC11522475): > *"We compared the B-ALL and AML patient-matched samples (designated **148752** and **152985**, respectively), and while 46 oncoprotein-binding sites were detectable in the B-ALL sample **at diagnosis**, treatment with the menin inhibitor reduced the genome-wide MLL N and C terminal signal to undetectable levels"* — and *"**prior to treatment** with the menin inhibitor … in the 148752 B-ALL sample, and **during treatment**, in the 152985 AML sample"*. | aliquot | state | |---|---| | **148752** | infant B-ALL **at diagnosis — UNTREATED** | | **152985** | the AML **during revumenib** | **So this is n = 1 patient with a paired pre/post design** — smaller than this entry first claimed and better controlled than its correction claimed. q143 read this correctly from the paper; q144 read it wrongly from GEO and amended its own pre-registration; **my brief asserted the GEO reading to both.** Third correction to this entry. 🔵 **And q144's positive control turns on it:** of 27 samples, **exactly one calls zero — 152985**, the treated AML. **Its own pre-treatment partner calls 345.** --- 🔴 **CORRECTED 2026-09-24 (cycle 124): THIS IS ONE PATIENT, NOT TWO.** `148752` (infant B-ALL) and `152985` (infant AML) are the **same infant before and after a lineage switch**. The GEO `!Sample_description` field says so explicitly — *"148752 … infant BALL lineage-switching sample treated with the Menin inhibitor **(paired with 152985)**"* and the reciprocal for `152985` — and I never parsed that field. **The infant menin-inhibitor occupancy arm is n = 1 PATIENT**, which is a much weaker resource than this entry claimed. **Both cycle-122 agents caught this independently**, from the same field, and the coordinator verified it against the raw record before applying it. The same correction applies to the KMT2A::AFF1 samples: `p279` and `p318` are **one** infant's B-ALL and its lineage-switched AML *(paired with p318 / paired with p279)*, so that arm is **two** patients and not three. 🟠 **And the compound is now known: revumenib**, named in the paper's abstract and verified there. The deposit itself says only *"Menin inhibitor"*, 60 times, and **neither the dose nor the exposure is recorded in the paper, the GEO record or the supplement.** --- **Statement.** `absence-no-menin-inhibitor-occupancy-dataset-in-an-infant-derived-kmt2a-aff1-line` holds that every chromatin measurement under a pharmacological menin inhibitor is in an adult-derived line. **GSE252378 contains the infant experiment.** Two **primary infant lineage-switching patient samples** — `148752` (infant B-ALL) and `152985` (infant AML) — were profiled **under a menin inhibitor** with **menin** (Bethyl A300-115A), **KMT2A-N1/N2/C1/C2**, **ENL** and **DOT1L** CUT&RUN, two replicates each [E1]. **And the absence survives, because both are `KMT2A-ENL`.** The index fusion is KMT2A::AFF1. This is the fourth time in this project that an absence has been narrowed rather than refuted by looking, and it is the most informative of the four: **the experiment is not hard, it is not unpublished, and it is not unfunded — it has simply not been done in this fusion.** That is a far stronger position from which to ask for it than *"nobody has tried"*. **The same deposit holds the untreated index-fusion arm.** Against the `KMT2A-AFF1` group label: **25 samples**, covering primary patient samples `p279`, `p318`, `p454` and the **SEM** cell line, with **AFF1** (two antibodies), **DOT1L**, **ENL** and **H3K4me3** CUT&RUN. **Every one is untreated.** So within a single deposit the field has the index fusion without the drug, and the drug without the index fusion — **the two halves of this project's central question, never crossed.** This is the same shape as cycle 115's error (two true margins, no join), except here the join genuinely is empty. **This deposit is cited by ONE entry in this corpus**, and was surfaced by the deposit index built this cycle rather than by any reading. **171 samples; raw data for 51 are withheld for donor privacy.** **Cite the paper, not the series.** The GEO series is titled *"KMT2A oncoproteins induce epigenetic resistance to targeted therapies"*; the publication is Janssens DH et al., **"MLL oncoprotein levels influence leukemia lineage identities"**, *Nat Commun* **15**:9341 (2024), **PMID 39472576**. An agent was caught by exactly this mismatch last cycle.

Go deeper
Why 'in the wrong fusion' is the useful form of this absence

The old form — no infant menin-inhibitor chromatin dataset exists — is an argument for generating one from scratch. The new form is an argument for one more arm on an existing, working protocol, by a group that has already run it in infant material, with the antibodies validated and the analysis pipeline published.

And it changes what the absence is evidence FOR. A missing experiment can mean the question is unasked, unaskable, or uninteresting. Here it is none of those: the question was asked, in infants, under the drug, with a menin antibody — and answered for KMT2A-ENL. The index-fusion gap is now a sampling accident rather than a frontier.

🔴 What this does NOT license. KMT2A-ENL and KMT2A::AFF1 are different fusions with different elongation-complex stoichiometry, and this project has already measured that between fusion backgrounds per-gene ordering transfers but magnitude does not. Reading the 148752/152985 result as a proxy for the index fusion would repeat the error that cost cycles 109 and 110.

CORRECTED: the fusion labels in this deposit are reliable after all

🔴 This section previously said GSE252378 carried a 42.9% metadata contradiction rate and that its group field was offset. Both claims were mine and both were wrong, retracted the same day they were written.

The cause was an off-by-one in my parser: GEO's brief format emits !Sample_title before !Sample_geo_accession, and my parser started each record at the accession, so every sample was compared against the next sample's title. Re-parsed with correct record boundaries:

claimedtrue
antibody disagreements9 of 21 (42.9%)0 of 44 (0.0%)
GSM8371030 titled SEM_AFF1_1group KMT2A-EPS15group KMT2A-AFF1

So the fusion assignments in this deposit are clean, and the counts this entry rests on are firmer than it originally claimed, not weaker. The 25 KMT2A-AFF1 samples, the 20 KMT2A-ENL menin-inhibitor samples and the two infant patient identities all stand on a field that agrees with its own titles throughout.

🔵 The one substantive caveat that survives is unrelated to metadata integrity: the characteristics say only "the Menin inhibitor" and never name the compound, dose or exposure, so this deposit still cannot be placed on the dose axis that cycles 115–117 built.

🟠 And the lesson runs the other way from the one this entry first drew. I read a defect into a deposit that did not have one, then used it to discount the deposit's own labels — which is the more dangerous direction of error, because it would have licensed ignoring exactly the evidence this entry was written to surface.

Facts and figures
questionhas menin occupancy under a menin inhibitor ever been measured in infant material?
answeryes - GSE252378, two primary infant lineage-switching patient samples, menin/KMT2A/ENL/DOT1L CUT&RUN under a menin inhibitor - but both are KMT2A-ENL, not KMT2A::AFF1
accessionGSE252378
total samples171
raw withheld for privacy51
menin inhibitor samples20
kmt2a aff1 samples are all untreatedtrue
the join that is emptythis deposit holds the index fusion WITHOUT the drug and the drug WITHOUT the index fusion; the cell that would answer this project's central question is empty
cited by n entries in this corpus1
how it was foundthe deposit index built this cycle, not by reading; a paper-only citation index cannot see a deposit
absence classmeasured-not-curated
bite backThis entry says the infant menin-inhibitor experiment is easy because someone has done it. If that is used to argue the index-fusion version is a quick add-on, note that neither the compound nor the dose nor the exposure is recorded in the metadata read here, and this project has spent three cycles establishing that dose is exactly where menin-inhibitor chromatin results stop being comparable.
group field counts
KMT2A-AFF21.000KMT2A-AFF2: 1KMT2A-EPS1510.0KMT2A-EPS15: 10KMT2A-ENL20.0KMT2A-ENL: 20KMT2A-AFF125.0KMT2A-AFF1: 25no group field115.0no group field: 115
publication
{
  "pmid": "39472576",
  "published_title": "MLL oncoprotein levels influence leukemia lineage identities",
  "geo_series_title": "KMT2A oncoproteins induce epigenetic resistance to targeted therapies.",
  "note": "the series title is NOT the published title; cite the paper"
}
menin inhibitor patients
{
  "148752": "infant B-ALL AT DIAGNOSIS - UNTREATED - and the pre-treatment baseline",
  "152985": "the AML of the same infant DURING revumenib",
  "CORRECTED_2026_09_24_cycle_128": "GEO's 'treated with the Menin inhibitor' is a CASE label on both aliquots; only 152985 is under the drug",
  "source": "the paper's own text, read at PMC11522475; verified by the coordinator",
  "consequence": "n=1 patient with a paired pre/post design, not two treated patients"
}
group field counts — the numbers behind the chart
{
  "KMT2A-AFF1": 25,
  "KMT2A-ENL": 20,
  "KMT2A-EPS15": 10,
  "KMT2A-AFF2": 1,
  "no group field": 115
}
kmt2a aff1 subjects
{
  "CORRECTED_2026_09_24": "p279 and p318 are ONE patient's B-ALL and lineage-switched AML",
  "patients": [
    "p279/p318 (one infant, two states)",
    "p454",
    "SEM (cell line)"
  ],
  "consequence": "two primary patients, not three"
}
kmt2a aff1 antibodies
[
  "AFF1 (PA5-77068)",
  "AFF1 (MBS190886)",
  "DOT1L",
  "ENL",
  "H3K4me3"
]
metadata offset
{
  "RETRACTED": "the claimed 42.9% antibody contradiction and the claimed group-field offset were both an off-by-one in the coordinator's parser",
  "true_antibody_disagreements": "0 of 44 (0.0%)",
  "GSM8371030_group": "KMT2A-AFF1, correctly matching its SEM_AFF1_1 title",
  "consequence": "the fusion assignments in this deposit are reliable"
}
compound
{
  "name": "revumenib",
  "source": "named in the abstract of PMID 39472576, verified",
  "dose": null,
  "exposure": null,
  "note": "the deposit says only 'Menin inhibitor'; no dose or exposure anywhere"
}
Still open
  1. Identify the compound, dose and exposure for the 148752 and 152985 menin-inhibitor arms by reading Janssens 2024 (PMID 39472576) directly. Without them this deposit cannot be placed on the dose axis cycles 115-117 built, and that axis is where menin-inhibitor chromatin results stop being comparable.
  2. Open the 25 untreated KMT2A-AFF1 libraries. AFF1, DOT1L, ENL and H3K4me3 CUT&RUN in three primary infant-fusion patient samples plus SEM, cited by one entry in this corpus, is the largest under-used index-fusion chromatin resource found so far.
  3. Determine which 51 samples have raw data withheld, and whether processed peak calls are available for them. If the menin-inhibitor arms are among the withheld, the reanalysis proposed above is not possible as stated.
  4. Ask whether the KMT2A-ENL menin result transfers. This project has measured that per-gene ordering transfers between fusion backgrounds while magnitude does not; testing that on these libraries against the index-fusion arm in the same deposit is a within-deposit comparison and therefore unusually clean.
Every source behind this entry (1)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    Parsing all 171 sample records of GEO series GSE252378: 20 samples carry 'treated with the Menin inhibitor' in their characteristics, all belonging to two primary infant lineage-switching patient samples, 148752 described as infant B-ALL and 152985 as infant AML, both carrying 'group: KMT2A-ENL' consistently across all 20 records. Their antibody panel by sample title covers Menin replicates 1 and 2, KMT2A-N1, KMT2A-N2, KMT2A-C1, KMT2A-C2, ENL replicates 1 and 2 and DOT1L replicates 1 and 2. Against the group field, 25 samples are labelled KMT2A-AFF1, 20 KMT2A-ENL, 10 KMT2A-EPS15 and 1 KMT2A-AFF2, with 115 carrying no group field; the KMT2A-AFF1 set covers primary samples p279, p318 and p454 and the SEM cell line with AFF1, DOT1L, ENL and H3K4me3 antibodies, and none of them carries a drug treatment. Comparing sample title against the sample's own characteristics on antibody identity, with record boundaries taken from !Sample_title rather than from !Sample_geo_accession, gives 0 disagreements in 44 comparable records; an earlier count of 9 in 21 was an off-by-one in the coordinator's parser and is retracted.

    GSE252378, Gene Expression Omnibus - all 171 sample records, parsed 2026-09-24 from the GEO accession endpoint · Janssens DH, Duran M, Otto DJ, Kirkey D, Yi JS, Mullighan CG and co-submitters; parsed by the Oncofusion knowledge graph coordinator · NCBI Gene Expression Omnibus 2024
    What this evidence cannot show

    THE COMPOUND, DOSE AND EXPOSURE ARE NOT NAMED. The characteristics say only 'treated with the Menin inhibitor', so this entry cannot say whether it is VTP-50469, revumenib or another, and the dose-matching argument built over cycles 115-117 cannot be applied to it. CORRECTED 2026-09-24: this caveat previously said the deposit's group field was offset and its counts approximate. That was an off-by-one in the coordinator's own parser - GEO emits !Sample_title before !Sample_geo_accession - and re-parsing gives 0 of 44 antibody disagreements with GSM8371030 correctly carrying group KMT2A-AFF1. The fusion assignments are reliable. NOTHING HERE OPENS A SINGLE LIBRARY: no signal was downloaded, no peak called, no occupancy computed, so this establishes what was deposited and not what it shows. RAW DATA FOR 51 OF THE 171 SAMPLES ARE WITHHELD for donor privacy, and which 51 was not determined. BOTH PATIENTS ARE LINEAGE-SWITCHING SAMPLES, a highly selected population.

    Evidence that argues against this (4)
    • The framing 'the absence survives' is the flattering reading of a discovery that this project's absence entry was less well-bounded than it claimed. That entry enumerated eight occupancy datasets under a menin inhibitor and asserted every one was adult-derived; GSE252378 is a ninth, it is infant, and it was cited in this corpus the whole time. The entry's conclusion happens to survive because the fusion differs, which is luck rather than rigour, and a reader should weight the enumeration method accordingly.
    • Both patients are LINEAGE-SWITCHING samples, which is a highly selected and unusual population - and lineage switch is itself concentrated in KMT2A-rearranged disease under immunotherapy pressure. Menin occupancy in a switching leukaemia may not represent menin occupancy in an ordinary infant KMT2A-r leukaemia at diagnosis, so even the KMT2A-ENL result these samples support is narrower than 'the infant experiment has been done'.
    • Calling the index-fusion gap 'a sampling accident rather than a frontier' is an inference about why the experiment was not run, and no evidence here supports it. The authors may have had a reason - availability of infant AFF1 material under treatment, antibody performance in that background, or a deliberate scope - and this entry converts silence into a claim about ease.
    • The correction to this evidence went in the reassuring direction - the data turned out cleaner than claimed - which is the direction that gets checked least. The specific numbers here (171 samples, 20 menin-inhibitor, 25 KMT2A-AFF1) were produced by the same coordinator script that was wrong about the offset, re-run after one fix, and have not been confirmed by any independent parse.
Connected to