Bounded with every archive and every route named: after both public SUV39H1 deposits were read this cycle, there is still NO SUV39H1 occupancy measurement in any haematopoietic cell, any KMT2A-rearranged cell, or any damage state - in ENCODE (three routes, zero), ChIP-Atlas (one deposit, MCF-7 only) or GEO (two deposits, glioma and breast)
also called the SUV39H1 lineage gap · what is still absent after Q114
Now that both public datasets have been read, the gap is easy to state: nobody has ever measured where this enzyme sits in a blood cell. Every measurement that exists comes from a brain tumour or a breast tumour, and the disease this project is about is a leukaemia.
The same thing, for a scientist
**What is absent, stated at the narrowest width it can be stated at.** No measurement of **SUV39H1 chromatin occupancy** exists in (a) any haematopoietic cell of any lineage or maturity, (b) any *KMT2A*-rearranged cell of any lineage, or (c) any damage or perturbed state of any cell - in any public archive, as of 2026-09-22 [E1][E2]. **The absence class is `never-measured`, and this time the word is earned, because the archive half was actually run** [E1]: | archive | route | holding | |---|---|---| | **ENCODE** | `/targets/SUV39H1-human/` | **HTTP 404** | | ENCODE | `type=Target&searchTerm=SUV39H1` | **total 0** | | ENCODE | `type=Experiment&target.label=SUV39H1` | **total 0** | | ENCODE | controls through the identical template | BARD1-human **200**, EXO1-human **404** | | **ChIP-Atlas** | `hg38/target/SUV39H1.5.tsv` | **4 experiments, SRX7030945-48, all MCF-7** = GSE139196 re-indexed | | **GEO** | Q113's `SUV39H1[Title] AND Homo sapiens[Organism] AND gsm[ETYP]` sweep, 75 GSMs read | **6 occupancy samples, 2 cell types**: SU-DIPG13/36 and MCF-7 | | **ReMap** | not retried this cycle | registry records **HTTP 401** - an access failure, not an absence | | **GTRD** | **not queried** | unknown | **Two of the eight rows are honest holes and are named as holes.** ReMap 401 is an access failure that the registry already records, and GTRD was never asked. So the bound is *"absent from ENCODE, ChIP-Atlas and GEO by the routes listed"*, not *"absent from the world"*. That distinction is the whole content of the registry row this beat was told to check, and the previous version of this absence - `absence-no-h3k9-writer-or-reader-localised-at-the-kmt2a-locus` - got it wrong in exactly this way by writing `never-measured` from an ENCODE-only census. **What is NOT absent any more, so that nobody re-runs it.** SUV39H1 occupancy at the *KMT2A* locus in a human cell: **measured this cycle**, in two DIPG lines, and negative. Writer-versus-mark coincidence in the same nuclei: **measured this cycle**, Spearman 0.148 and 0.078. The 90 kb gene body: **measured**, 0.94x and 1.11x the random-tile mean. **Cost of closing it: one CUT&RUN.** `exp-q114-a-suv39h1-cutandrun-in-a-cd34-progenitor-and-a-kmt2a-rearranged-line`, `cost_tier: med`. This is a **never-measured** absence in the cheapest class - the reagent exists, has a deposited human track, and CUT&RUN works on 10^5 primary cells.
Why this absence is load-bearing and not bookkeeping
mark-h3k9me3 argues that H3K9 methylation restrains the fusion programme, so lowering it is dangerous. mechanism-h3k9me1-2-ctcf-balance-gates-kmt2a-rearrangement argues that H3K9me1/2 promotes rearrangement, so lowering it is protective. The two point in opposite therapeutic directions, and the enzymes in the middle of that dispute have never been placed at the locus in a cell of the right lineage.
Until this cycle the SUV39H1 side of that could not even be asked from an archive. Now it can be asked, and the answer in two non-haematopoietic systems is that the enzyme is not there. What that licenses is a narrower dispute, not a resolved one: if SUV39H1 is not at the locus in a blood cell either, then the H3K9me3 this graph argues about must come from SETDB1, from EHMT2/G9a, or from nowhere - and of those three, SETDB1 has been read at this locus in K562 only and EHMT2 has never been read here at all.
| script | no script - archive routes and their response codes recorded inline; the companion reads are analyses/q114-suv39h1-at-the-kmt2a-locus/ |
|---|---|
| status | run |
| run date | 2026-09-22 |
| run by | Q114-suv39h1-at-the-locus |
| absence class | never-measured |
| what is absent | Any measurement of SUV39H1 chromatin occupancy in a haematopoietic cell of any lineage, in a KMT2A-rearranged cell of any lineage, or in any damage or perturbed state - in ENCODE, ChIP-Atlas or GEO. |
| what is NOT absent | SUV39H1 occupancy at the KMT2A locus in a human cell (measured this cycle in two DIPG lines, negative); writer-versus-mark coincidence in the same nuclei (measured, Spearman 0.148 and 0.078); the 90 kb gene body (measured, 0.94x and 1.11x the random-tile mean); a condition-contrasted SUV39H1 occupancy dataset (GSE139196, +/- estradiol 40 min). |
| name list NOT used | a KMT1A[Title] sweep of GEO DataSets has still not been run by any beat |
| cost to close | one CUT&RUN with CST #8729S - exp-q114-a-suv39h1-cutandrun-in-a-cd34-progenitor-and-a-kmt2a-rearranged-line, cost_tier med |
| shown for fusion | NO - nobody has looked in any KMT2A fusion system |
archives queried
{
"ENCODE": "three routes, all empty, with positive and negative controls through the identical URL template",
"ChIP-Atlas": "one file, four MCF-7 experiments, which are GSE139196 re-indexed",
"GEO": "Q113's SUV39H1[Title] sweep over 75 GSMs, plus both deposits re-read per sample"
}archives NOT queried or failed
{
"ReMap": "HTTP 401 per the existing registry row; NOT retried this cycle",
"GTRD": "not queried at all"
}name lists used
[ "SUV39H1", "KMT1A (present in GSE139196's own sample titles)" ]
- An absence this cheap to close will be cited as a reason to be cautious about H3K9-directed agents, which is the right direction but the wrong warrant: caution should rest on the measurement nobody has made, not on the fact that nobody has made it.
- Does a KMT1A[Title] sweep of GEO DataSets return a deposit the SUV39H1[Title] sweep missed? GSE139196's own sample titles use KMT1A, so the alias is demonstrably in live use in deposit metadata, and this is a one-query check that no beat has run.
- Does ReMap hold SUV39H1 behind its 401? The registry records the access failure and nobody has retried it or asked the maintainers. An access failure recorded as an absence is exactly the defect this graph keeps finding in its own instruments.
- Is there a SUV39H1 occupancy dataset in a mouse haematopoietic cell? This beat and Q113 both bounded on Homo sapiens only. A murine HSPC track would not transfer, but it would tell this graph whether the experiment is technically routine or has never been made to work.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
Three ENCODE routes return nothing for SUV39H1 and ChIP-Atlas's whole hg38 SUV39H1 holding is the four MCF-7 samples of GSE139196, queried 2026-09-22 with controls through the identical URL template.
ENCODE Portal SUV39H1 target and experiment queries; ChIP-Atlas hg38 target table SUV39H1.5.tsv (26 genes, columns SRX7030945-48, all MCF-7) · ENCODE Project Consortium; Oki S, Ohta T et al. · ENCODE Portal / ChIP-Atlas (DBCLS) 2026What this evidence cannot show
Dated and route-scoped. ReMap was not retried (the registry records a 401) and GTRD was not queried, so two archives are unexamined and are named as unexamined.
Evidence that argues against this (1)
- A target-label query cannot see an experiment filed under an unrelated label, and the registry's own row on tagged targets exists because of that. The union of the exact-label and searchTerm routes narrows but does not eliminate it. source
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
The only two human SUV39H1 occupancy deposits in GEO are in patient-derived H3K27M glioma and in MCF-7; neither is haematopoietic, neither is KMT2A-rearranged, and both are resting-state.
GSE225028 (SU-DIPG13, SU-DIPG36 CUT&RUN) and GSE139196 (MCF-7 ChIP-seq +/- estradiol) per-sample SOFT records · Xin DE et al.; Yang F et al. · NCBI GEO 2023What this evidence cannot show
Rests on Q113's GEO title sweep (SUV39H1[Title], 75 GSMs read) plus this beat's re-reading of both deposits. A deposit whose sample titles name only a cell line, or which uses the retired symbol KMT1A in the title, would be invisible to that sweep - Q113 records this limitation itself and a KMT1A[Title] GEO sweep has still not been run.
Evidence that argues against this (1)
- GSE139196's own sample titles read 'ChIP on SUV39H1 (KMT1A)', which is the clearest possible demonstration that the alias is in live use in deposit metadata. That the title sweep caught this one on the approved symbol is luck, not coverage. source
the experiment that closes this absence
the lineage gap that survives after both deposits were read
replaces that entry's ENCODE-only 'never-measured' with an archive-named absence about LINEAGE, not about existence
the census that bounds it
the writer in this mark's dispute has never been placed in a cell of the right lineage