HOXA is occupied as a 107 kb cluster; IRX1 is occupied as one gene, with both IRX clusters cold
When the leukaemia protein switches on HOXA, it lights up the whole neighbourhood of genes at once. When it switches on IRX1, it lights up that one gene and nothing around it.
The same thing, for a scientist
In the HOXA-high patients the entire **107 kb HOXA cluster** (HOXA1 txStart to HOXA13 txEnd) carries KMT2A-N at **2.65-3.23x** mean promoter — a coordinated, cluster-wide occupancy. The **IRXA cluster** (1.72 Mb, IRX4-IRX2-IRX1) never exceeds **0.29x** in any patient and the **IRXB cluster** (1.05 Mb, IRX3-IRX5-IRX6) never exceeds **0.25x**, while the IRX1 gene body alone reaches **6.35x** [E1]. The two homeobox programmes therefore have different chromatin architectures, and reading them as symmetric alternatives understates the difference.
Measured at matched scales
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.
| patient | HOXA cluster (107 kb) | IRXA cluster (1.72 Mb) | IRXB cluster (1.05 Mb) | IRX1 gene (5.6 kb) |
|---|---|---|---|---|
| p179 | 3.23 | 0.022 | 0.052 | 0.02 |
| p247 | 2.81 | 0.027 | 0.052 | 0.07 |
| TB051939 | 2.65 | 0.089 | 0.118 | 0.22 |
| p279 | 0.63 | 0.053 | 0.048 | 6.08 |
| TB920488 | 0.42 | 0.289 | 0.252 | 0.59 |
| p454 | 0.16 | 0.116 | 0.044 | 6.35 |
| p318 | 0.11 | 0.063 | 0.060 | 2.68 |
| TB986292 | 0.06 | 0.048 | 0.072 | 2.38 |
Every coordinate was read from the downloaded UCSC refGene table 2.
This is why the U9 scale check mattered
The pre-registration required every comparison at both promoter and cluster scale, so that a verdict tracking the window size would be caught as a fact about the ruler. The IRX1-vs-HOXA ordering is identical at 2 kb, 8 kb, gene body and 20 kb, and the Spearman strengthens from -0.571 to -0.714 as the window widens to the gene. It collapses only at the 1.72 Mb cluster window — and the collapse is explained: that window is ~99.7% intergenic. So U9 does not fire, and the collapse is a measurement of focality rather than a threshold artefact.
| hoxa cluster bp | 107097 |
|---|---|
| irxa cluster bp | 1723977 |
| irxb cluster bp | 1047453 |
| irx1 gene bp | 5572 |
| hoxa cluster max | 3.231 |
| irxa cluster max | 0.289 |
| irxb cluster max | 0.252 |
| irx1 gene max | 6.35 |
- A focal single-gene dependency is in principle easier to disrupt than a coordinated cluster, but it is also easier to bypass by switching to a paralogue - and IRX2 is already up in one of these patients.
- Is the IRX1 focality a property of the locus or of how recently it was switched on? A cluster-wide spread might develop over time and these are single timepoints.
- Does the fusion reach IRX1 through a CpG island, as the KMT2A CXXC address model predicts? The CpG content of the IRX1 promoter was not measured in this beat.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
The 107 kb HOXA cluster carries KMT2A-N at 2.65-3.23x in HOXA-high patients, while neither the 1.72 Mb IRXA nor the 1.05 Mb IRXB cluster exceeds 0.29x in any patient despite the IRX1 gene body reaching 6.35x.
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
Cluster means over multi-megabase windows are dominated by intergenic sequence and are not directly comparable to a gene-body mean; the comparison that carries the claim is HOXA cluster vs IRXA cluster, which are both mostly intergenic, and there the contrast is ~30-100x. 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
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
All interval coordinates (HOXA1, HOXA13, IRX1-IRX6) were read from the UCSC hg38 refGene table downloaded this cycle and passed an automated overlap/chromosome/ordering audit before any signal was scored.
UCSC hg38 refGene table (coordinates for every interval used in q145) · UCSC Genome Browser · UCSC 2026What this evidence cannot show
The audit checks chromosome, ordering and gene overlap; it cannot check that the chosen transcript is the biologically relevant one, and for multi-transcript genes the longest NM_ was taken by rule rather than by curation.
Evidence that argues against this (1)
- A gene whose longest NM_ transcript has an unusual alternative first exon would get a TSS that no expressed isoform uses; IRX1 has only one NM_ transcript so this does not affect the primary locus. UCSC Genome Browser 2026
the whole 107 kb HOXA cluster carries oncoprotein at 2.65-3.23x
- supports →IRX1 — Iroquois homeobox 1, the gene that names the HOXA-negative half of infant t(4;11) ALL
IRX1 is focal: both IRX clusters stay below 0.29x while the gene reaches 6.35x
HOXA10 rises with HOXA9 as part of the coordinated cluster