What is left after the missing arm was read: there is no H3K9me1, me2 or me3 measurement at the KMT2A breakpoint region in ANY haematopoietic cell, and the founding paper's own archive contains none because all 60 of its samples are epithelial
also called the absence that survived cycle 59 · 60 samples, one cell type · the H3K9 brake has never been seen in blood
The experiment nobody had looked at has now been read, and it answered its question. But every sample in that archive came from eye cells. Nobody has ever measured these chemical tags on the KMT2A gene in a blood cell, which is the only place the disease happens.
The same thing, for a scientist
`absence-h3k9me3-at-the-kmt2a-locus-was-assayed-and-never-reported` is now **closed for the cell type it concerned**: the eight unreported H3K9me3 samples were read, the me1/me2 positive controls passed, and H3K9me3 does not follow the KDM3B-G9a balance at the *KMT2A* CTCF anchor in RPE [E1]. **The absence that replaces it is larger and is not closable by compute.** All **60** samples in GSE210480 are hTERT-RPE [E2]. So every chromatin measurement supporting the rearrangement brake - the baseline H3K9me1 depletion at the BCR, the rise on KDM3B depletion, the CTCF loss at exon 11, and now the me3 null - exists in **one epithelial line and nowhere else**, while the cytogenetic phenotype it is supposed to explain was demonstrated in U937, KG1a, a primary AML line, a primary AML organoid, cord-blood CD34+ HSPCs and mouse spleen [E2]. **Absence class: never-measured, in a haematopoietic cell, for all three H3K9 methyl states at this locus.** It is not a literature gap that a better query would close, it is not a tooling gap, and it is not a resolution gap - it is a **cell-type** gap, and closing it needs wet work. One cheap partial step exists and was not taken by this beat: public CD34+ histone panels in ENCODE and BLUEPRINT could answer the **baseline** half - is H3K9me1 depleted at the *KMT2A* BCR in a progenitor as it is in RPE? - with no wet work at all.
What was closed, what was not, and the exact shape of what is left
| question | status after cycle 59 |
|---|---|
| Does H3K9me3 change at the KMT2A CTCF anchor on KDM3B or G9a depletion? | ANSWERED in RPE: no (1.01x, 0.98x; every arm inside the control replicate range) |
| Does H3K9me3 change over the wider BCR? | ANSWERED in RPE: modestly and in both single knockdowns equally, so not by the balance (1.34x, 1.38x); a quarter of it is a chromosome-wide shift |
| Does H3K9me1 reproduce the published BCR result? | ANSWERED in RPE: yes, 3.74x at the element, BCR rank 72/18,082 chr11 bins |
| Does CTCF fall at the anchor? | ANSWERED in RPE: yes, 0.30x at canonical exon 11, and at none of eleven other intervals |
| Is Gray's 'exon 11 CTCF site' the same element this graph adjudicated? | ANSWERED: yes, centroid within 34 bp, 3.8 kb from the DCAL cut point |
| Any H3K9me1/2/3 measurement at this locus in a haematopoietic cell | NEVER MEASURED |
| Any CTCF-occupancy change at this locus under KDM3B depletion in a haematopoietic cell | NEVER MEASURED |
| Any of the above in a fetal progenitor or a B-cell precursor | NEVER MEASURED |
| Any sequenced rearrangement junction from this mechanism, in any cell | NEVER MEASURED - 'break apart' is FISH probe separation throughout |
How the archive bound was established. All 60 GSM records of GSE210480 were retrieved as text and every one carries !Sample_organism_ch1 = Homo sapiens with RPE as the source; the series contains no haematopoietic sample of any kind 2. The build was confirmed independently from the bigWig chromosome table rather than from GEO's genome_build field, because in a sibling series three human HCT116 samples are annotated Mus musculus and the error propagates to esummary db=gds - so a metadata-based cell-type or organism filter is not a route this project trusts 1.
Why this absence is differently shaped from its predecessor. The predecessor was measured-not-analysed: the assay existed, the files existed, the answer was one compute job away, and cycle 57 costed it precisely enough that cycle 59 closed it in an afternoon. This one is never-measured. There is no file to read. The archive was checked and the archive does not contain it.
The cheapest partial step, which this beat did not take
Do the public-data baseline check first. ENCODE and BLUEPRINT both carry histone-modification panels for CD34+ haematopoietic progenitors. If H3K9me1 or H3K9me2 tracks exist for CD34+ cells, then the baseline half of the question - is H3K9me1 depleted over the KMT2A BCR relative to its flanks in a progenitor, as it is in RPE (0.54x versus 1.28x)? - is answerable in an afternoon with the tooling this beat has already proven, at 200 bp resolution or better if the tracks allow.
It matters which way that comes out. The baseline depletion is the first link in the published chain and the only one that is a property of the resting cell rather than of a perturbation. If H3K9me1 is not depleted at the BCR in a CD34+ progenitor, then the chain's starting condition does not exist in the cell type where the disease starts, and the whole mechanism becomes an epithelial phenomenon - which would be the single most important negative available anywhere in this beat's territory, obtainable for free. If it is depleted, the perturbation experiment (exp-h3k9me3-and-ctcf-at-the-kmt2a-anchor-in-a-cd34-progenitor) becomes much better motivated.
This is named as an unrun step rather than done, because it is a different search - an archive survey across ENCODE and BLUEPRINT for a cell type and a mark, not a re-read of one series - and doing it badly is worse than flagging it. It is the first thing the next cycle should do in this territory, and it is cheaper than everything else on the list.
| absence class | never-measured |
|---|---|
| what is absent | any H3K9me1, H3K9me2 or H3K9me3 measurement at the KMT2A breakpoint cluster region or its exon-11 CTCF anchor in any haematopoietic cell, and any CTCF-occupancy change at that anchor under KDM3B or G9a depletion in any haematopoietic cell |
| what replaced | absence-h3k9me3-at-the-kmt2a-locus-was-assayed-and-never-reported, which was measured-not-analysed and is now closed for RPE at 1 kb resolution |
| archive check run | true |
| archive queried | NCBI GEO, all 60 GSM records of GSE210480 retrieved as text - the founding paper's own deposit, which is the matching archive for an occupancy question. NOT queried: ENCODE and BLUEPRINT CD34+ histone panels, which is the named weak point of this bound. |
| not a literature gap | no query would close it. Not a tooling gap - the tooling works and was used. Not a resolution gap - that is recorded separately. It is a CELL-TYPE gap and it needs wet work, except for the baseline half which may be public. |
| cheapest partial step unrun | survey ENCODE and BLUEPRINT for H3K9me1 or H3K9me2 tracks in CD34+ haematopoietic progenitors and ask whether H3K9me1 is depleted over the KMT2A BCR relative to its flanks, as it is in RPE (0.54x versus 1.28x). An afternoon with tooling already proven. It tests the FIRST link of the published chain - the only link that is a property of the resting cell - and a negative would be the most important negative available in this territory. |
| what closes it fully | exp-h3k9me3-and-ctcf-at-the-kmt2a-anchor-in-a-cd34-progenitor - cord-blood CD34+ HSPCs, control and KDM3B knockdown, ChIP for H3K9me1/2/3 and CTCF at the anchor with ZNF274 and GAPDH as mark-identity controls. Medium cost, weeks, no animals, no cohort. |
| cost tier | med |
| the deeper gap it sits inside | 'break apart' is FISH probe separation everywhere in the founding paper and there is no sequenced junction in any cell type, so even a complete haematopoietic chromatin dataset would join a chromatin state to a cytogenetic proxy rather than to a rearrangement |
| localization | chromatin at chr11q23.3, the KMT2A breakpoint cluster region and its exon-11 CTCF anchor - in a nucleus that has never been a haematopoietic one for this measurement |
| subnuclear localization | also never measured at this element in any cell type: no imaging, Hi-C or nuclear-compartment assay places the KMT2A anchor relative to the lamina or a nuclear speckle, which is notable because the one paper cited for CTCF protein loss under stress localises that loss to SC-35 speckles |
| cell type | ABSENT - cord-blood CD34+ HSPCs, a fetal liver progenitor, a B-cell precursor or a KMT2A-rearranged blast are the cell types in which this has never been measured. The 60 samples that exist are all hTERT-immortalised retinal pigment epithelium. |
| cell state | ABSENT for every state that matters: resting progenitor, cytokine-stimulated progenitor, and progenitor under topoisomerase II poison. The measurements that exist are a 72 h siRNA knockdown in proliferating epithelium. |
| epigenetic readout | H3K9me1, H3K9me2 and H3K9me3 at the KMT2A BCR - measured in epithelium, never in blood; and CTCF occupancy at the exon-11 anchor under KDM3B depletion, likewise |
| shown for fusion | Not shown for any fusion, and that is part of the absence: the mechanism concerns the intact KMT2A locus before a rearrangement exists, so the cell type that matters is a normal haematopoietic progenitor rather than a fusion-bearing blast - and no H3K9 measurement at this locus exists in either. |
- Replacing a closed absence with a bigger one can read as moving the goalposts. It is not: the closed question was 'what does the unreported arm say', and it now has an answer. The new entry exists because the answer's scope turned out to be one cell type, and saying so is the alternative to letting an RPE null be quoted as a general fact - which is exactly what the predecessor entry's own bite-back warned would happen.
- The cheapest step named here - the public CD34+ baseline check - was not run by this beat, and an unrun cheap step inside a bounded absence is a task pretending to be a finding. It is flagged as unrun in three places for that reason, and the next cycle should do it before anything else in this territory.
- Do ENCODE or BLUEPRINT carry H3K9me1 or H3K9me2 in CD34+ progenitors? This is the named weak point of this entry's bound and the cheapest next step in the whole beat. If such a track exists, the baseline half of this absence is already closed in public data.
- Is the baseline H3K9me1 depletion at the KMT2A BCR reproduced in any non-RPE cell at all? It is the first link of the chain and the only one that is a property of a resting cell rather than a perturbation, so it is the cheapest link to test and the most damaging to lose.
- Has anyone ever measured H3K9 methylation at the KMT2A locus in a KMT2A-rearranged blast? That is a different question again from the progenitor one - it asks about the locus in the disease rather than about its genesis - and this beat did not search for it.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
The eight unreported H3K9me3 samples of GSE210480 were read and answered their question in RPE - H3K9me3 is 1.01x of control at the KMT2A CTCF anchor on KDM3B depletion and 0.98x on G9a depletion with every arm inside the control replicate range, against H3K9me1 at 3.74x - and the build was verified from the bigWig chromosome table (hg19 plus exactly 100,000 bp on chr1, chr2, chr11 and chrX) rather than from GEO metadata.
GEO GSE210480 H3K9me1/2/3 and CTCF bigWigs read with pybigtools 0.3.0, 2026-09-22; pre-registration, scripts and full output at analyses/q108-gse210480-h3k9me3-kmt2a-bcr/ · Gray ZH, Chakraborty D, Duttweiler RR, Alekbaeva GD, Murphy SE, Chetal K, Ji F, Ferman BI, Honer MA, Wang Z, Myers C, Sun R, Kaniskan HU, Toma MM, Bondarenko EA, Santoro JN, Miranda C, Dillingham ME, Tang R, Gozani O, Jin J, Skorski T, Duy C, Lee H, Sadreyev RI, Whetstine JR · NCBI GEO (data accompanying Cell 186:4528-4545.e18) 2023What this evidence cannot show
n = 2 per arm, no spike-in, histone tracks 1 kb binned, and the result is a null under those conditions - 'no change detectable', not 'no change'. The closure being claimed is therefore narrow: the RPE question is answered to the precision the deposited files support, and the entry that posed it should be amended rather than deleted.
Evidence that argues against this (1)
- Calling the predecessor absence 'closed' could itself be premature: the deposited files are 1 kb binned and a sharply localised me3 change at the CTCF motif would be averaged into invisibility, so the RPE question is answered at 1 kb and not at 200 bp. The raw reads in SRA would settle it, and `absence-gse210480-processed-tracks-are-1-kb-binned` deliberately recommends NOT spending that compute while the cell-type gap is open - a judgement that could be wrong. Iqbal Lab oncofusion knowledge graph 2026
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
All 60 samples of GSE210480 are human hTERT-RPE cells with no haematopoietic sample of any kind, while the cytogenetic phenotype the series' chromatin is meant to explain was demonstrated by the same paper in U937, KG1a, a primary AML line, a primary AML organoid, cord-blood CD34+ HSPCs and mouse spleen - so no H3K9me1, H3K9me2 or H3K9me3 measurement at the KMT2A breakpoint region exists in any blood cell.
NCBI GEO series GSE210480, all 60 GSM records retrieved as text 2026-09-22 (acc.cgi?acc=GSE210480&targ=gsm&form=text&view=brief) - every !Sample_organism_ch1 is Homo sapiens and every source is RPE · NCBI Gene Expression Omnibus 2026What this evidence cannot show
This bounds ONE series. It establishes that the founding paper's own deposit contains no haematopoietic chromatin; it does NOT establish that no such measurement exists anywhere, which would need an archive-wide survey of H3K9me1/2/3 tracks in haematopoietic cells that this beat did not run. ENCODE and BLUEPRINT both carry CD34+ histone panels and were NOT checked - so the BASELINE half of the question may already be publicly answerable, and the never-measured claim is properly scoped to the PERTURBATION (KDM3B or G9a depletion in a blood cell), which certainly does not exist. The cell-system inventory for the paper's cytogenetic arm is taken from this graph's mechanism entry rather than re-read from the figures.
Evidence that argues against this (1)
- The never-measured claim is bounded by one series plus this graph's reading of one paper, and the honest weak point is named in the caveat: ENCODE and BLUEPRINT CD34+ histone panels were not checked. If an H3K9me1 track in CD34+ cells exists, part of this absence is already closed in public data and the entry overstates the gap. That check is listed as the cheapest next step for exactly this reason. Iqbal Lab oncofusion knowledge graph 2026
replaces it: the measured-not-analysed absence is closed for RPE and a never-measured cell-type absence takes its place
the experiment that would close it; the public CD34+ baseline check would close half of it for free
every chromatin link of that chain exists in one epithelial line and nowhere else
the resolution gap is recorded separately; this one is a cell-type gap and needs wet work
the cell-type gap is why the RPE null cannot be read as a safety finding
the mark's absence in a blood cell (Q108) and the enzymes' absence in every cell (this beat) are two halves of the same hole