Oncofusion

start / graph / ana-the-entire-public-record-holds-one-menin-chip-under-a-dot1l-inhibitor-and-it-is-the-one-q136-used

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

Searching 32 GEO series and 2,312 samples for a menin readout under DOT1L inhibition: exactly ONE distinct experiment exists, MOLM13_MEN1-ChIP_EPZ5676_4d at 100 nM, and it is the sample Q136 already used - so the deflationary reading of Q136 is blocked not by a missing assay but by a ~50-fold DOSE GAP, and Q136's drug-class conclusion rests on a single unreplicated sample in a single cell line

also called one menin ChIP under a DOT1L inhibitor, worldwide · the dose gap, not the assay gap · Q136 rests on n=1 · 2,312 samples, one joint experiment

A recent result here said a menin-targeting drug and DOT1L-targeting drugs do very different things to menin on DNA. The obvious way to attack that result is to ask whether the DOT1L drugs would look the same if given at the much higher doses used elsewhere. Searching every relevant public dataset shows that the measurement needed exists exactly once, at the low dose, and it is the very measurement the original result used. Nobody has ever measured menin under a high dose of a DOT1L drug, so the objection cannot be tested - and the original result turns out to rest on one unrepeated experiment.

The same thing, for a scientist

**Statement.** Q136 found that on the MEN1 channel the drug classes separate completely — **VTP-50469 leaves 0 of 19 menin peaks at canonical targets; EPZ5676 leaves 14, compound 10 leaves 14, compound 11 leaves 8**, permutation p = 0.0003 — and named its own deflationary reading: *if menin loss also appears at the 5 µM / 72 h DOT1L exposures the MLL1-channel work used, the channel distinction collapses into a dose distinction.* Cycle 115 found GSE260742 has **no MENIN samples at all** and recorded that finding a dataset was worth more than another beat on that deposit. **Searched: 32 GEO series matching `DOT1L AND menin`, 2,312 samples, scored at SAMPLE level as a cross-tabulation rather than as two margins.** **`D4` — neither pre-registered branch, and the question was misposed.** A menin readout under a DOT1L inhibitor **does** exist. There is **exactly one distinct experiment of it in the public record**: `MOLM13_MEN1-ChIP_EPZ5676_4d`, deposited twice (GSE149183 and its SuperSeries GSE149185). **It is the sample Q136 already used, at 100 nM for 4 days.** **Nothing anywhere carries a menin readout at the ~5 µM exposures the MLL1-channel work uses — a dose gap of roughly 50-fold.** So the deflationary reading is not blocked by a missing assay; it is blocked by dose, and **it remains untestable on existing data.** **And the same count bounds the result it was meant to attack.** Q136's drug-class conclusion for EPZ5676 rests on **one unreplicated sample, one cell line, one dose, one timepoint.** The compound-10 and compound-11 arms are additional samples but from the same deposit, lab and experiment. **That is a far thinner base than "permutation p = 0.0003" conveys** — the p-value describes the peak-level null, not the number of independent experiments. **The margins would have misled, which is the point of scoring the join.** 14 series contain a menin sample somewhere and 18 contain a DOT1L-inhibitor sample somewhere, and several contain both — **but in different samples.** Two cycles ago this project briefed an agent on exactly that error.

Go deeper
The by-product is worth more than a negative: a deposit cited 48 times and never opened

The search surfaced GSE173599 — "Chromatin state changes induced by targeted degradation of MLL-AF9 are phenocopied by combined DOT1L and MENIN inhibition", 75 samples, PMID 35245435 verified as Olsen, Godfrey et al., Molecular Cell 2022, PMC9044330.

That paper is cited in 48 files of this corpus. Its data is mentioned in zero.

What it contains, and this project has never had any of it:

  • a dTAG-VHL degron of MLL-AF9 at 500 nM, 3 h / 24 h / 4 d — removing the fusion outright, the reference point every eviction claim here currently lacks;
  • ChIP under that degron of MEN1, MLL1, DOT1L, AFF4, CDK9, RNA Pol II (total, pS2, pS5), H3K27ac, H3K4me3, H3K79me2, H3K9ac — the whole complex plus the elongation machinery;
  • combination arms: EPZ5676 1 µM with VTP-50469 100 nM, on H3K27ac and MLL1, in both MOLM13 and HCB1.

The combination arm is the closest thing in existence to the experiment cycle 83's threat calls for, and the degron arm is the control this thread has been reasoning without for thirty cycles.

My own first pass got this wrong, and the script is why it was caught

I hand-inspected two series — GSE173599 and GSE172153 — found no joint sample in either, and concluded "zero samples anywhere combine a menin ChIP with a DOT1L inhibitor." False. Scripting the same test across all 32 series found the joint sample immediately, in the deposit this project had already analysed.

Two samples is not a search, and the failure mode is specific: I picked the two series that looked most promising from their titles, which is selection on the thing being measured. The scripted version also recorded the margins it would have reported, so the cross-tab discipline is visible in the artefact rather than only in the prose.

Facts and figures
questiondoes any public deposit measure menin under DOT1L inhibition, and at what dose?
series searched32
samples scanned2312
distinct joint experiments1
joint samples at or above 1uM0
the comparison doseSGC0946 5 uM / 72 h, used by the MLL1-channel work
dose gap~50-fold
what this does to Q136its EPZ5676 drug-class arm rests on one unreplicated sample, one cell line, one dose, one timepoint; the permutation p describes the peak-level null, not the number of independent experiments
my own errorI hand-checked two series, found no joint sample, and concluded none existed anywhere; the scripted search over all 32 found it in the deposit this project had already analysed. Two samples is not a search.
absence classmeasured-not-curated
bite backThis entry tells the project that a central recent result rests on n=1 and that its main objection is untestable. Both readings discourage work: the first invites discarding Q136, the second invites dropping the objection. The constructive reading is that GSE173599's combination arm exists and has never been opened, and if that turns out not to bear on the question, this entry will have spent a cycle's attention on a survey.
the joint sample
{
  "title": "MOLM13_MEN1-ChIP_EPZ5676_4d",
  "deposits": [
    "GSE149183",
    "GSE149185 (SuperSeries)"
  ],
  "dose": "100 nM EPZ5676, 4 days",
  "already_used_by": "Q136"
}
margins would have said
{
  "series_with_a_menin_sample": 14,
  "series_with_a_dot1l_inhibitor_sample": 18,
  "note": "several contain both but in DIFFERENT samples - the join is the test"
}
the by product
{
  "accession": "GSE173599",
  "samples": 75,
  "pmid": "35245435",
  "paper": "Olsen, Godfrey et al., Molecular Cell 2022, PMC9044330",
  "cited_in_corpus_files": 48,
  "accession_mentioned_in_corpus_files": 0,
  "contains": [
    "dTAG-VHL degron of MLL-AF9 at 500 nM, 3h/24h/4d",
    "ChIP of MEN1, MLL1, DOT1L, AFF4, CDK9, RNA Pol II total/pS2/pS5",
    "H3K27ac, H3K4me3, H3K79me2, H3K9ac",
    "combination arms EPZ5676 1 uM + VTP-50469 100 nM in MOLM13 and HCB1"
  ]
}
Still open
  1. Open GSE173599. Its combination arm (EPZ5676 1 uM + VTP-50469 100 nM, MLL1 and H3K27ac, two cell lines) is the closest existing experiment to cycle 83's threat, and its degron arm is the remove-the-fusion reference point this thread has never had.
  2. Does the degron arm's MEN1 ChIP at 3 h answer a different and better question - what happens to menin when the fusion is removed rather than when a drug is applied? That is the specificity control no menin-inhibitor experiment can provide.
  3. Re-run this search with the DOT1L inhibitor named by structure or by target rather than by the four trade names, and without requiring the series to mention menin. U3 bounds the current answer to the terms used.
  4. Is 48 citations with 0 data uses a general pattern here? If the corpus routinely cites conclusions without opening deposits, the same audit over every heavily-cited PMID would find more unopened data.
Every source behind this entry (1)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.analysis

    A GEO DataSets search for 'DOT1L AND menin AND gse[ETYP]' returns 32 series; fetching every sample title gives 2,312 samples. Scoring each sample for both a menin readout and a DOT1L-inhibitor treatment yields exactly one distinct joint experiment, MOLM13_MEN1-ChIP_EPZ5676_4d, appearing in GSE149183 and its SuperSeries GSE149185. No joint sample anywhere carries a stated dose at or above 1 uM. Fourteen series contain a menin sample and eighteen a DOT1L-inhibitor sample without any single sample being both. The search also returned GSE173599, 75 samples, whose publication PMID 35245435 resolves to Olsen, Godfrey et al., Molecular Cell 2022, and which is cited in 48 files of this corpus while its accession appears in none; it contains a dTAG-VHL degron of MLL-AF9 at 500 nM with ChIP of MEN1, MLL1, DOT1L, AFF4, CDK9, RNA Pol II, H3K27ac, H3K4me3, H3K79me2 and H3K9ac, and combination arms of EPZ5676 1 uM with VTP-50469 100 nM.

    What this evidence cannot show

    THE SEARCH IS ONLY AS WIDE AS ITS TERMS. It covers EPZ5676, EPZ004777, SGC0946 and pinometostat via the GEO DataSets index; a deposit naming its DOT1L inhibitor some other way, or not indexed under both 'DOT1L' and 'menin', is invisible, so every absence here means 'not findable by this search'. SAMPLE TITLES ARE NOT PROTOCOLS: a title naming MEN1 may be a knockout rather than a ChIP - GSE172153's menin samples are sgMEN1 guides - and a title omitting a dose does not mean no dose was used, so the '50-fold dose gap' rests on doses STATED IN TITLES plus the series metadata Q136 read, not on protocols. The claim that GSE173599's data is untouched is a grep for the accession across this repository and would miss an entry that used the data while citing only the paper. Nothing here re-analyses any of these deposits; this is a survey.

    Evidence that argues against this (3)
    • Calling Q136's base 'one unreplicated sample' is the harshest available framing and is not the only fair one. Its conclusion rests on nineteen canonical loci scored against a strength-matched permutation null, with a second library (DOT1L ChIP) agreeing and an internal positive control; and the compound-10 and compound-11 arms are genuinely separate treatments even if from one experiment. 'One sample' is true of the EPZ5676 arm specifically and understates the design.
    • The dose-gap framing assumes the MLL1-channel comparison at 5 uM / 72 h is the right thing to match, but that exposure comes from a different deposit, cell system and antibody. Matching the dose would not make the experiments comparable, and a menin ChIP at 5 uM in MOLM13 would still not be the experiment cycle 83's threat describes. The gap this entry names is real and its closure would be less decisive than the framing implies.
    • That GSE173599's accession appears in no file while its paper appears in 48 is presented as an indictment and may simply be normal practice: a corpus can cite a paper's conclusions without needing its raw data, and this project's chromatin thread has been deliberately narrow. The finding is only damning if the data would have changed something - which this entry asserts and does not demonstrate.
Connected to