Bounded absence: no published study has tested whether a KMT2A fusion GAINS chromatin sites under a menin inhibitor - and the same pipelines found gains for two other proteins when they looked
also called nobody looked for gains · the missing gained-peak analysis · one-sided differential analysis
If the drug pushed the fusion protein onto new places on the DNA, we would only know if somebody had looked for new places, and nobody has. Every study asked what was lost - and the two that did ask about gains, for other proteins, both found them.
The same thing, for a scientist
**The null this absence sits against, stated first.** The field's position, and this graph's, is that a menin inhibitor removes the fusion from chromatin and that the fusion is then degraded or diluted away. That is a coherent model, it is what the class's abstracts say, and the one human sample profiled under revumenib is consistent with it [E4]. **This entry does not claim relocation happens.** It claims that the observation which would distinguish relocation from removal has never been made, and quantifies how completely it has never been made. **Query bounding.** Against a 21-term menin-inhibitor name list in the correct one-tag-per-term form (rule 94), run 2026-09-21 and with querytranslation recorded in `data`: | conjunct | records | on topic | |---|---:|---:| | `redistribut*[tiab]` | 1 | **0** - a gastric-cancer nanoparticle biodistribution paper | | `relocaliz*[tiab]` | 0 | 0 | | `retarget*[tiab]` | 0 | 0 | | `"new binding sites"[tiab]` | 0 | 0 | | `"gain of occupancy"[tiab]` | 0 | 0 | | `chromatin[tiab]` (the enumeration base) | 61 | 8 occupancy datasets | **Operand controls, because a zero from a query that cannot return a one is worthless (rule 67).** `redistribut*[tiab] AND chromatin[tiab]` returns **856**; `menin[tiab] AND occupancy[tiab]` returns **26**; `menin[tiab] AND redistribut*[tiab]` returns **4**. The instrument works. Note separately that `"gain of occupancy"[tiab]` returns **0 across the whole of PubMed**, so that phrase is not a usable probe for anything and its zero here carries no information - recorded so that nobody re-runs it and counts it twice. **Full-text bounding, because a PubMed zero is bounded by abstracts (rule 80).** The gained-peak analysis would live in a supplementary figure, not a title. I therefore read the full texts of the four papers most likely to contain one - Krivtsov 2019, Perner 2023, Soto-Feliciano 2023 and Kerry 2017 - and checked their methods, figure legends and supplementary-table titles [E1][E2][E3]. Krivtsov ships *Table S4: list of genes that lose occupancy >2-fold of MLL1 and Menin* and no gain equivalent [E1]. Perner called MACS2 peaks in every condition and reported TSS-window losses only [E2]. **Soto-Feliciano did run the gain analysis - and found more than 13,000 new UTX sites under MI-503** [E3]; the 2026 JUND paper likewise reports increased JUND chromatin occupancy under the drug [E5]. **So the absence is not technical and it is not a limitation of the assay. Two of eight datasets asked the gain question, both got an answer, and neither asked it of the oncoprotein.** **Rule-of-three bound.** 0 of 8 occupancy datasets tested the fusion for gains, so with 95% confidence fewer than 37.5% of such datasets do - a weak bound, honestly reported, and the informative statement is not the proportion but the enumeration, which names all eight.
What would close this
Three things, in increasing cost:
ana-reanalysis-of-ml2-mll1n-chip-for-gained-peaks-under-vtp-50469- the data are public in GSE127508 and include a fusion-only line. One node-day of compute.exp-resistant-versus-sensitive-fusion-fate-men1-interface-mutant- the isogenic MEN1 interface mutant supplies the control that a treated-only peak set most needs.exp-calibrated-early-timecourse-fusion-profiling-under-menin-inhibition- the definitive version, with spike-in calibration, epitope attribution and sub-24-hour timepoints.
If (1) returns nothing, this absence should be rewritten as a bounded negative - no persistent gained fusion occupancy at day 3 in three cell lines - which is a real result and a much stronger statement than the current one.
Why a one-sided analysis is the default, and why that is the whole story here
Nothing hid this. A differential ChIP analysis is usually specified around a hypothesis, the hypothesis was the drug removes the oncoprotein from its targets, and a tornado plot ordered by control signal answers that hypothesis completely while making the complementary question invisible - the rows that were empty before treatment are compressed at the bottom of the plot and contribute nothing to the eye or to the metagene. Add a 2-fold loss threshold and a target-gene list defined in the untreated condition, and gains cannot enter the result by any route.
This is the same shape as the defects this project has documented in its own instruments: a measurement whose design presupposes the answer's direction. The remedy is identical - call peaks in the treated condition too, and report the union.
| script | none - bounded PubMed query plus full-text reading, 2026-09-21 |
|---|---|
| status | run |
| query form note | one tag per term throughout, per rule 94; no parenthesised group carries a [tiab] tag |
| querytranslation chromatin base | ("revumenib"[Title/Abstract] OR "ziftomenib"[Title/Abstract] OR "bleximenib"[Title/Abstract] OR "enzomenib"[Title/Abstract] OR "VTP-50469"[Title/Abstract] OR "SNDX-5613"[Title/Abstract] OR "KO-539"[Title/Abstract] OR "MI-503"[Title/Abstract] OR "MI-3454"[Title/Abstract] OR "MI-2-2"[Title/Abstract] OR "MI-463"[Title/Abstract] OR "MI-538"[Title/Abstract] OR "JNJ-75276617"[Title/Abstract] OR "DSP-5336"[Title/Abstract] OR "DS-1594"[Title/Abstract] OR "BMF-219"[Title/Abstract] OR "menin inhibitor"[Title/Abstract] OR "menin inhibitors"[Title/Abstract] OR "menin inhibition"[Title/Abstract] OR "menin-MLL inhibitor"[Title/Abstract] OR "menin-MLL inhibitors"[Title/Abstract]) AND "chromatin"[Title/Abstract] |
| denominator | 8 occupancy datasets enumerated in ana-what-menin-inhibition-has-actually-been-measured-on-chromatin |
| numerator testing fusion gains | 0 |
| rule of three bound | fewer than 37.5% of such datasets test fusion gains, 95% confidence, n=8 |
| caveat scope | zero in titles and abstracts; supporting figures searched only in the five full texts listed |
| shown for fusion | no - the gain question has not been asked of any KMT2A fusion in any system |
| absence class | measured-not-analysed |
| absence class basis | GSE127508 holds paired ML2 MLL1n ChIP under DMSO and VTP; every published pipeline tabulated losses only. |
| absence class assigned | 2026-09-21 (cycle 47, king) |
counts — the numbers behind the chart
{
"drug_list AND chromatin": 61,
"drug_list AND redistribut*": 1,
"drug_list AND relocaliz*": 0,
"drug_list AND retarget*": 0,
"drug_list AND \"new binding sites\"": 0,
"drug_list AND \"gain of occupancy\"": 0
}operand controls — the numbers behind the chart
{
"redistribut*[tiab] AND chromatin[tiab]": 856,
"menin[tiab] AND occupancy[tiab]": 26,
"menin[tiab] AND redistribut*[tiab]": 4,
"\"gain of occupancy\"[tiab] alone": 0
}off topic hits
{
"42449441": "hyaluronic-acid liposome/mesoporous-silica nanoplatform co-delivering BMF-219 and 5-FU in gastric cancer; 'redistribution' is drug biodistribution"
}full texts read
[ "PMC7227117 (Krivtsov 2019)", "PMC10157896 (Perner 2023)", "PMC9827117 (Soto-Feliciano 2023)", "PMC5263239 (Kerry 2017)", "PMC6726985 (Wu 2019)" ]
not read
[ "PMID 38234854 bioRxiv, KMT2A oncoproteins induce epigenetic resistance to targeted therapies - highest-priority unread item for this absence" ]
raw data availability
{
"raw_reads_confirmed_available": true,
"checked": "2026-09-21 (king)",
"accession": "GSE127508 (PMID 31821784), SuperSeries of GSE127506, GSE127507, GSE137813, GSE137814",
"evidence": "GEO SOFT family file lists 78 samples and 79 SRA experiment ids. SRX5444450 verified live in SRA: GSM3635632, MOLM13_RNA_DMSO_d2_REP1, Illumina, 6,230,066 spots, 467,254,950 bases, Homo sapiens.",
"route_that_works": "https://ftp.ncbi.nlm.nih.gov/geo/series/GSE127nnn/GSE127508/soft/GSE127508_family.soft.gz -- the SOFT family file, which is authoritative and is not behind bot detection.",
"routes_that_gave_a_FALSE_NEGATIVE": [
"elink dbfrom=gds db=sra on the SuperSeries UID returns NO LINK; SuperSeries do not carry sample-level SRA links, their subseries do.",
"esearch db=sra term=PRJNA524881[BioProject] returns 0, while the SRX ids under that same BioProject resolve individually. The field query is a false zero.",
"Fetching the GEO acc.cgi HTML returned a reCAPTCHA challenge page, which parsed cleanly as '0 subseries, no SRA' -- a bot-detection page is not a negative result."
],
"consequence": "The rDNA re-map and any gained-peak reanalysis CAN be done from raw reads for this dataset. An earlier read of this record as coverage-tracks-only (TDF/WIG) was wrong; those are merely the supplementary files listed on the SuperSeries."
}- An unasked question is weaker evidence than a negative result, and this entry must not be cited as though the fusion has been shown not to relocate. The honest statement is that the class's mechanism of action is specified in a direction nobody has tested.
- If a gain exists and is where resistance comes from, every combination now in trial was designed against an incomplete model, and the relevant sites are ones no current agent targets.
- Conversely, if the reanalysis shows no gain, this absence should be retired promptly rather than left standing - a stale absence in this project has previously suppressed real work for seven cycles.
- Does PMID 38234854 (bioRxiv, KMT2A oncoproteins induce epigenetic resistance to targeted therapies) contain a gained-peak analysis? Not read for this beat and the single most likely counter-example.
- Do any of the deposited processed peak files in GSE127508 or its sibling accessions already contain treated-condition peak calls that would answer this without realignment?
- Is there an equivalent absence in the DOT1L-inhibitor literature - has anyone looked for gained fusion occupancy after pinometostat? If neither class has been asked, the gap is a habit of the field rather than a fact about menin.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The pivotal VTP-50469 study quantified occupancy as TSS-anchored window signal, applied a greater-than-2-fold loss threshold, and shipped a supplementary table titled as a list of genes that lose occupancy more than 2-fold of MLL1 and menin, with no corresponding gain table; its own text reports that MLL1 was not globally displaced, so there was retained signal available to redistribute.
A Menin-MLL Inhibitor Induces Specific Chromatin Changes and Eradicates Disease in Models of MLL-Rearranged Leukemia. · Krivtsov AV, Evans K, Gadrey JY, et al. · Cancer Cell 2019What this evidence cannot show
Read from the full text of PMC7227117 on 2026-09-21, including methods, figure legends and supplementary-item titles; the supplementary tables themselves were not downloaded, so a gain column inside Table S2 cannot be excluded from the text alone. Three adult-derived human lines, one drug concentration, one timepoint.
Evidence that argues against this (1)
- Absence of a gain table in the manuscript is not absence of the analysis - authors routinely run and discard analyses that show nothing. If the gained-peak analysis was run and was empty, this absence entry overstates the gap, and only the authors or a reanalysis can settle that. Krivtsov AV et al. 2019
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The clinical-resistance study called MACS2 peaks with matched input controls in every condition and reported only TSS-window occupancy losses for menin and MLL1, with no differential peak analysis in either direction.
MEN1 mutations mediate clinical resistance to menin inhibition. · Perner F, Stein EM, Wenge DV, et al. · Nature 2023What this evidence cannot show
Read from the full text of PMC10157896. The statistical test used on the occupancy data is a two-sided Mann-Whitney on TSS signal distributions, which is not direction-blind in practice because it was applied to MLL1-target gene sets defined in the untreated state.
Evidence that argues against this (1)
- Because peaks were called in every condition, a gained-peak list may exist in this study's processing outputs even though it is not reported, which would make the absence analytic rather than experimental for this dataset too. Krivtsov AV et al. 2019
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The same class of pipeline detects gains when asked: MI-503 treatment produced a substantial increase in UTX chromatin association with more than 13,000 additional binding sites beyond baseline, predominantly at promoters, not explained by increased UTX expression or protein stability.
A Molecular Switch between Mammalian MLL Complexes Dictates Response to Menin-MLL Inhibition. · Soto-Feliciano YM, Sánchez-Rivera FJ, Perner F, et al. · Cancer Discov 2023What this evidence cannot show
Murine MLL-AF9 leukaemia cells with the tool compound MI-503, promoter-centric analysis (TSS +/- 2 kb), and the 13,000 figure is described in the body text from supplementary panels. UTX begins from very few peaks, so its dynamic range for a gain is unusually favourable.
Evidence that argues against this (1)
- UTX's gain is mechanistically expected - the paper's thesis is that MLL1-menin restricts MLL3/4-UTX access - so it demonstrates the pipeline can call gains but not that anyone would have thought to call them for the fusion, which was expected to leave. Krivtsov AV et al. 2019
- E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The one human sample profiled under revumenib showed oncoprotein and menin reduced to undetectable levels while ENL persisted at target loci, which is consistent with the null - removal rather than relocation - and is the only in-patient evidence on either side.
MLL oncoprotein levels influence leukemia lineage identities. · Janssens DH, Duran M, Otto DJ, et al. · Nat Commun 2024What this evidence cannot show
n = 1, confounded by a concurrent lymphoid-to-myeloid lineage switch that independently reduces oncoprotein levels; an undetectable CUT&RUN signal conflates loss of binding with loss of protein, and no genome-wide gain analysis of the oncoprotein under drug is reported for that sample either.
Evidence that argues against this (1)
- If the oncoprotein protein level itself fell, that sample cannot discriminate eviction from relocation at all - a protein that is not present cannot be anywhere, and the observation reduces to a statement about abundance. Wu Y et al. 2019
- E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
A second gain under menin inhibition is reported for JUND: menin-MLL inhibitor treatment increased JUND chromatin occupancy, upregulating target genes and contributing to resistance.
Menin-MLL inhibitors enhance JUND activity in MLLr leukemic cells contributing to tumorigenesis and therapy resistance. · Ozyerli-Goknar E, Nizamuddin S, Fetsch V, et al. · Blood 2026What this evidence cannot show
Chromatin binding assays are described at abstract level with no stated genome-wide differential method, cells, concentration or timepoint; JUND competes directly for the MLL1/2 pocket on menin, so its liberation is a special case.
Evidence that argues against this (1)
- Two gains in two attempts is a base rate of one, computed on n = 2, and both gaining proteins had a specific prior reason to gain. It is a reason to look at the fusion, not a prediction that the fusion will gain. Soto-Feliciano YM et al. 2023
One of only two absences in the corpus that names a deposited accession (GSE127508) — cited here as the positive example of the measured-not-analysed class, where closing the gap is compute rather than wet work.
SURVIVES at full text (28 records, 0 on topic) and the class upgrades: GSE260742 holds FLAG-MLLAF9 ChIP-seq under VTP50469 and the gain call was never made
the question this absence records as untested has now been tested in the human index fusion, and the answer is a small reproducible gain rather than none
the enumeration is the denominator for the bounded absence
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
the gain question has never been asked of the index fusion
bounds what is known about this drug's effect on fusion localisation
the analysis that would close this absence from public data
the experiment that would close it definitively
two of the four documented gains show the pipeline detects gains when asked
establishes that the missing fusion analysis is a choice of question, not a limit of the assay