In drug-naive primary infant KMT2A::MLLT1 B-ALL, menin is as enriched as the fusion N-terminus at the top fusion promoters (2.73x vs 3.12x over an accessibility reference) while the wild-type C-terminal reporter sits at background (1.10x)
In a baby's leukaemia cells taken before any menin-blocking drug, menin is sitting exactly where the fusion protein is - and it is not sitting where the normal, unbroken version of the same gene is. That is the clearest direct evidence this project has that the menin the drug targets is the menin attached to the fusion.
The same thing, for a scientist
Using a filter-free promoter ranking (no peak caller, no target list) in sample 148752 - the diagnostic, pre-revumenib aliquot of the lineage-switching infant - and dividing every channel's enrichment by an out-of-sample IgG reference to remove the accessibility gradient: at the **top 50** KMT2A-N promoters the accessibility-corrected enrichment is **menin 2.73x, KMT2A-N 3.12x, ENL 11.76x, DOT1L 16.10x, and KMT2A-C 1.10x**. The C-terminal (wild-type) channel is **at background** at the very promoters where the N-terminal channel and menin are most enriched. The independent gene-set control agrees: the KMT2A-N ranking puts 40 of its top 50 on the authors' published oncoprotein-target list and the menin ranking puts 14 of 50, while the KMT2A-C ranking puts **1 of 50 (p = 0.84)**. Menin's position tracks the channel that contains the fusion and not the channel that contains only wild-type KMT2A.
The ladder, and where it stops meaning anything
Accessibility-corrected enrichment (channel fold over its own genome-wide promoter median, divided by the same quantity for IgG; 1.0 = no enrichment beyond what a non-specific antibody sees) 1:
| k | menin | KMT2A-N | KMT2A-C | ENL | DOT1L |
|---|---|---|---|---|---|
| 50 | 2.73 | 3.12 | 1.10 | 11.76 | 16.10 |
| 200 | 1.64 | 1.64 | 1.10 | 2.76 | 4.76 |
| 1000 | 1.26 | 1.32 | 1.02 | 1.70 | 2.88 |
| 5000 | 1.09 | 1.19 | 1.01 | 1.36 | 2.26 |
The verdict does not flip across the ladder - menin tracks KMT2A-N at every k and KMT2A-C stays at 1.0-1.1 throughout - but the effect is concentrated at the top, as it must be: by k = 5000 the 'fusion promoter' set is a quarter of the genome's promoters and nothing is enriched in it.
The raw numbers, before any correction, so both readings are visible
Without dividing by IgG, the fold over each channel's own genome-wide promoter median at the top 50 is menin 14.41x, KMT2A-N 16.45x, KMT2A-C 5.78x, ENL 62.02x, DOT1L 84.88x - and IgG itself is 5.27x. So the uncorrected KMT2A-C figure of 5.78x, which reads as a real signal, is the accessibility gradient almost exactly. This is the case the IgG division was pre-registered for, and the IgG available is out-of-sample (see the caveat below), so the corrected column is a declared approximation and the raw column is given beside it rather than replaced by it.
The N/C contrast under decile matching, where it looks different and why
A second, stricter cut of the same question splits the top 10% of the KMT2A-N ranking into N-biased promoters (N rank far above C rank) and N&C-concordant ones, then matches the two groups on the N channel's own decile so that a self-stratified gradient cannot manufacture a difference (405 vs 405 promoters). Raw medians give menin 0.743 at N-biased versus concordant - menin lower where the fusion is enriched. But IgG gives 0.504 on the same contrast, i.e. N-biased promoters are substantially less accessible, and after that correction menin is 1.47x higher at N-biased promoters, against KMT2A-N's own 1.99x, ENL 2.05x, DOT1L 1.53x and KMT2A-C 1.26x. The sign of this contrast depends entirely on whether an accessibility reference is used, which is stated here rather than resolved, and it is why the top-k ladder rather than this contrast is the primary measure. The N-biased promoter list is nonetheless recognisable: PAX5, SENP6, CDK6, GUCY1A1, SKIDA1, MEIS1, MBNL1, HOXA7, ZEB2, HOXA9.
What this is a result about
Primary human infant B-ALL, KMT2A::MLLT1, one patient, one aliquot, before revumenib. It is not the index fusion KMT2A::AFF1, and this project has measured that per-gene ordering transfers between fusion backgrounds while magnitude does not - so 2.73x is a KMT2A::MLLT1 number and only the ordering should be carried across. It is also a lineage-switching patient, a selected population.
| sample | 148752, diagnostic infant B-ALL, KMT2A::MLLT1, before revumenib |
|---|---|
| measure | accessibility-corrected promoter enrichment = (channel fold over its own genome-wide promoter median at top-k) / (same for out-of-sample IgG) |
| fusion | KMT2A::MLLT1 - NOT the index fusion; carry the ordering, not the magnitude |
| shown for fusion | KMT2A::MLLT1 in one primary infant patient; nothing in KMT2A::AFF1 |
| epigenetic readout | none - this is protein occupancy, not a histone mark |
| status | run |
| script | analyses/menin-under-revumenib-in-an-infant-patient-pair/run.py |
| preregistration | analyses/menin-under-revumenib-in-an-infant-patient-pair/PREREGISTRATION.md |
top50 — the numbers behind the chart
{
"MENIN": 2.73,
"KMT2A_N": 3.12,
"KMT2A_C": 1.1,
"ENL": 11.76,
"DOT1L": 16.1
}top200 — the numbers behind the chart
{
"MENIN": 1.64,
"KMT2A_N": 1.64,
"KMT2A_C": 1.1,
"ENL": 2.76,
"DOT1L": 4.76
}top1000 — the numbers behind the chart
{
"MENIN": 1.26,
"KMT2A_N": 1.32,
"KMT2A_C": 1.02,
"ENL": 1.7,
"DOT1L": 2.88
}top5000 — the numbers behind the chart
{
"MENIN": 1.09,
"KMT2A_N": 1.19,
"KMT2A_C": 1.01,
"ENL": 1.36,
"DOT1L": 2.26
}uncorrected top50 fold over own median — the numbers behind the chart
{
"MENIN": 14.41,
"KMT2A_N": 16.45,
"KMT2A_C": 5.78,
"ENL": 62.02,
"DOT1L": 84.88,
"IgG": 5.27
}decile matched N biased vs concordant
{
"raw": {
"MENIN": 0.743,
"KMT2A_N": 1.005,
"KMT2A_C": 0.637,
"ENL": 1.033,
"DOT1L": 0.771,
"IgG": 0.504
},
"after_dividing_by_IgG": {
"MENIN": 1.47,
"KMT2A_N": 1.99,
"KMT2A_C": 1.26,
"ENL": 2.05,
"DOT1L": 1.53
},
"n_matched": [
405,
405
],
"sign_depends_on_accessibility_correction": true
}N biased promoters head
[ "PAX5", "SENP6", "CDK6", "GUCY1A1", "SKIDA1", "MEIS1", "MBNL1", "HOXA7", "ZEB2", "HOXA9" ]
- This is the evidence that the menin a menin inhibitor removes is fusion-associated menin, in a patient. It is one patient of a selected kind, in the wrong fusion for this project's purposes, and it rests on an accessibility reference borrowed from a different child.
- Menin's enrichment here (2.73x) is well below ENL's (11.76x) and DOT1L's (16.10x) at the same promoters. If occupancy magnitude predicts therapeutic leverage, menin is not the most occupied node of this complex at fusion targets - which is an argument for combination rather than for menin monotherapy.
- Does the same N-versus-C separation hold in the untreated KMT2A::AFF1 samples of this deposit (p279, p454, SEM)? That is the index-fusion version and it was not run here.
- How much of the menin at these promoters is LEDGF-delivered rather than fusion-delivered? No LEDGF/PSIP1 channel exists in this deposit.
- Why does the decile-matched N-biased contrast change sign on the accessibility correction, and would an in-sample IgG have resolved it? No in-sample IgG exists.
- Are N-biased promoters genuinely less accessible, or is the out-of-sample IgG reporting the other patient's accessibility rather than this one's?
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
In drug-naive infant KMT2A::MLLT1 B-ALL, accessibility-corrected promoter enrichment at the top 50 KMT2A-N sites is menin 2.73x and KMT2A-N 3.12x, while the wild-type-reporting KMT2A-C channel is 1.10x.
analyses/menin-under-revumenib-in-an-infant-patient-pair - pre-registered re-analysis of GSE252378 samples 148752 and 152985 · Q143-menin-under-drug-in-infants (Oncofusion Knowledge Graph) · this repository 2026What this evidence cannot show
n = 1 patient, one aliquot. The accessibility reference is an out-of-sample IgG from a different patient with a different fusion (KMT2A-EPS15), CUT&RUN only, so it is not a background control for the CUT&Tag menin channel at all. Menin is CUT&Tag and KMT2A-N is CUT&RUN, run three weeks apart, so the menin-versus-N comparison crosses assay and batch. No spike-in exists, so every figure is a within-library share.
Evidence that argues against this (2)
- Menin is delivered to MLL target genes by LEDGF/PSIP1 as well as by MLL itself, so menin at a fusion promoter does not by itself establish that the fusion is what put it there. Yokoyama A et al. 2008
- Menin also occupies promoters as part of wild-type MLL1/MLL2 complexes and in a fusion-independent role in several tissues, so co-localisation with the fusion channel is necessary but not sufficient for a fusion-dependent tether. Soto-Feliciano YM et al. 2023
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
The same ranking's independent gene-set control separates the two ends sharply: KMT2A-N recovers 40 of 50 published oncoprotein-target genes, menin 14 of 50, KMT2A-C 1 of 50 (p = 0.84).
analyses/menin-under-revumenib-in-an-infant-patient-pair - pre-registered re-analysis of GSE252378 samples 148752 and 152985 · Q143-menin-under-drug-in-infants (Oncofusion Knowledge Graph) · this repository 2026What this evidence cannot show
The target-gene set is the authors' own, derived partly from this patient's panel, so it is an independent definition rather than an independent replication.
Evidence that argues against this (1)
- A C-terminal epitope can be masked rather than absent at a busy promoter, so a C channel at background is consistent with the fusion interpretation without proving it. Olsen SN et al. 2022
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The paper independently reports 46 detectable oncoprotein-binding sites in this sample at diagnosis, which is the authors' own statement that the fusion is measurably on chromatin in 148752.
MLL oncoprotein levels influence leukemia lineage identities · Janssens DH, Duran M, Otto DJ, Wu W, Xu Y, Kirkey D, Mullighan CG, Yi JS, Meshinchi S, Sarthy JF, Ahmad K, Henikoff S · Nature Communications 15:9341 2024What this evidence cannot show
Their 46 is a peak-call count under their own SEACR threshold; my measure is a signal ranking with no peak call, so the two numbers are not comparable and only the direction is shared.
Evidence that argues against this (1)
- The authors describe the lineage-switching samples as having REDUCED oncoprotein levels relative to non-switching KMT2A-r leukaemias, so this patient's fusion occupancy is at the low end of the disease even in the diagnostic sample. Janssens DH et al. 2024
menin occupies the top fusion-N promoters at 2.73x accessibility-corrected enrichment in primary infant KMT2A::MLLT1
- related to →KMT2A::MLLT1 (MLL-ENL), t(11;19)(q23;p13.3)
the fusion whose N-terminal channel menin tracks
the N/C readout the result is built on
the positive control that licenses the ranking
the accessibility correction uses an out-of-sample IgG
the menin-versus-N comparison crosses assay and batch
DOT1L is the most enriched channel at these promoters (16.10x)
ENL is enriched 11.76x at the same promoters
the same measurement in the drug-free aliquot of the same patient
the positive half
the menin-tracks-N observation
confirms that beat's reading of 148752 as the drug-naive diagnostic aliquot against the deposit's own string