p279 carries the oncoprotein at HOXA AND at IRX1 — the published mutual exclusivity is not mutual exclusivity at chromatin
The textbook says infant leukaemias use either HOXA or IRX1, never both. One of these babies' leukaemias had the driver protein sitting on both at once.
The same thing, for a scientist
The iALL-HOXA9 / iALL-IRX1 split is defined by **mutually exclusive transcript expression** [E1]. In p279 — an infant KMT2A::AFF1 B-ALL at diagnosis — KMT2A-N occupies **HOXA9/HOXA10 at 2.92x** (20 kb window) **and IRX1 at 6.08x** (gene body), both far above the IgG floor of zero, with both N antibodies agreeing (IRX1 3.47 / 3.57; HOXA9 4.63 / 2.23) [E2]. This is the single clearest datum in the beat against reading the expression dichotomy as an occupancy dichotomy, and it was **not** predicted by the pre-registration.
It is not a depth artefact and not an antibody artefact
p279 is the deepest pair of N libraries in the set (promoter zero-rates 31.5% and 6.0%; P(a promoter is zero in both by chance) = 0.019). Both N antibodies — Millipore 05-764 mouse and CST D2M7U rabbit, which are different reagents and not replicates — put IRX1 at 3.47x and 3.57x and HOXA9 at 4.63x and 2.23x. The two reagents agree in sign at both loci, so U2 does not fire 2.
Three readings, and occupancy cannot choose between them
- A mixed blast population. p279 contains both an iALL-HOXA9 clone and an iALL-IRX1 clone, and bulk CUT&RUN averages them. This is the reading I favour on prior grounds and it is untestable in bulk data.
- Genuine co-occupancy in one nucleus. Then mutual exclusivity is a property of transcript output, not of binding, and the two are decoupled at this locus.
- p279 is transitional. It is the diagnosis sample of the one patient in the cohort who went on to lineage-switch, and its switched counterpart p318 has lost HOXA (0.23x) while keeping IRX1 (2.68x). p279 may be caught mid-transition from a HOXA to an IRX1 program.
Reading 3 is attractive and is the one I distrust most, because it is a narrative that fits n=1 and the sample was selected for having switched.
| patient | p279 |
|---|---|
| hoxa9 hoxa10 20kb | 2.924 |
| irx1 gene | 6.076 |
| irx1 promoter N1 | 3.468 |
| irx1 promoter N2 | 3.569 |
| hoxa9 promoter N1 | 4.63 |
| hoxa9 promoter N2 | 2.23 |
| igg at irx1 | 0 |
- If subgroup assignment is a mixture rather than a clone-level property, then treating a patient as 'the IRX1 subgroup' selects therapy against the fraction of their disease, and the untargeted fraction is the relapse.
- Single-cell CUT&Tag or a clonality-resolved assay on p279 would settle mixture-versus-co-occupancy, and it is the single most decisive cheap experiment this beat generates.
- Do any of the published iALL cohorts contain a patient expressing both HOXA9 and IRX1? Symeonidou's per-patient scatter would answer it and was not re-derived here.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The iALL-HOXA9 / iALL-IRX1 subgroups are defined by statistically significant mutually exclusive expression of HOXA9/HOXA10 and IRX1 (Spearman inverse correlation plus Fisher's exact test).
HOXA9/IRX1 expression pattern defines two subgroups of infant MLL-AF4-driven acute lymphoblastic leukemia · Symeonidou V, Ottersbach K · Exp Hematol 2021What this evidence cannot show
Mutual exclusivity was established on transcript abundance across two cohorts. It is a population-level statistical statement about cohorts, not a claim that no individual patient can express both.
Evidence that argues against this (1)
- Symeonidou herself notes one of the genes examined was 'not uniformly upregulated' in its subgroup, so the exclusivity was never absolute even in expression. Symeonidou V et al. 2021
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
In p279, KMT2A-N occupies HOXA9/HOXA10 at 2.92x and IRX1 at 6.08x mean promoter, with both N antibodies agreeing in sign at both loci and an IgG floor of zero at IRX1.
q145 re-analysis of GSE252378: IRX1 and HOXA oncoprotein occupancy across seven KMT2A::AFF1 patients (analyses/q145-irx1-in-hoxa-negative-infants) · Iqbal Lab oncofusion knowledge graph, beat q145 · internal analysis over deposited GSE252378 bigWigs 2026What this evidence cannot show
n = 1 patient, and it is the one patient in the cohort selected for having undergone a lineage switch, so it is not a random draw from infant t(4;11). Bulk CUT&RUN cannot distinguish co-occupancy in one nucleus from a mixture of two clones. Every number is a WITHIN-SAMPLE share: an interval's mean coverage divided by that same library's mean promoter coverage, so 1.0 = an average promoter in that library. **GSE252378 has no spike-in and no drug arm**, so no between-sample absolute statement is licensed; the only cross-sample quantity used is a ratio of two loci measured inside one library, where the unknown per-library scale cancels.
Evidence that argues against this (1)
- One deposit, one lab, one assay, one genome build, and activity tracks in only three of the eight sample records. A second cohort profiled by a different group could reverse any of this. Janssens DH et al. 2024
the clearest single datum that the expression dichotomy is not an occupancy dichotomy
- related to →IRX1 — Iroquois homeobox 1, the gene that names the HOXA-negative half of infant t(4;11) ALL
one patient carries the oncoprotein at IRX1 and at HOXA simultaneously
the experiment that would resolve mixture versus co-occupancy
- ← related top279 and p318: one infant, two diseases, and an oncoprotein target map that does not carry over
the same diagnostic sample's HOXA/IRX1 occupancy state