What is left of the resection absence after the compute was done: the resting state is now MEASURED and negative in 13 ENCODE datasets, so the whole question collapses onto the DAMAGE state, which is never-measured for all six proteins in every cell type on Earth
also called the damage-state gap · what Q35 could not close · the resection absence after cycle 63 · scope: the six named proteins only - NOT RPA (RPA at the 11q23 hotspot after etoposide: PMID 21048951)
Last cycle this graph said nobody had looked for repair proteins on the KMT2A gene, and that for two of them public data could answer it. We ran that analysis and the answer is no. What remains cannot be answered from any existing dataset, because nobody has ever measured these proteins at this gene in a cell that has actually been damaged.
The same thing, for a scientist
**This entry exists to stop the resection absence being re-quoted in its cycle-62 form.** Q32 recorded six proteins as never localised at the *KMT2A* locus and split the absence by class, with **BRCA1 and CBX3 as `measured-not-analysed`** - closable by compute. Q35 ran that compute over all 13 released ENCODE experiments (BRCA1 7, CBX3 4, and CBX1 2 added), with a pre-registered design, three nulls and a dynamic-range gate. See `ana-q35-brca1-cbx3-and-cbx1-are-not-at-the-kmt2a-breakpoint-cluster-in-any-encode-cell`. **The `measured-not-analysed` class is now discharged, and the answer was negative.** Zero called peaks in 13 of 13; the breakpoint cluster core at 0.57-1.24x the genome-wide median of same-width tiles; the exon-11 CTCF element below the median CTCF anchor of its own cell type in 10 of 13 tracks. **So here is the honest remaining shape of the absence, per protein, after the compute.** | protein | resting state, ENCODE | damage state, anywhere | what is left | |---|---|---|---| | **BRCA1** | **MEASURED, negative**, 7 experiments incl. K562 + GM12878 | **never-measured** | a damage-state ChIP | | **CBX3** (HP1-gamma) | **MEASURED, negative**, 4 experiments (3 usable) | **never-measured** | a damage-state ChIP | | **CBX1** (HP1-beta) | **MEASURED, negative**, 2 experiments | **never-measured** | a damage-state ChIP | | **KDM5A** | measured-not-analysed in A549/H1/HepG2; **no blood cell exists** | never-measured | a blood-cell ChIP first | | **BARD1** | ~~no ENCODE target record~~ **the target record EXISTS (200); no released TF ChIP-seq** - corrected 2026-09-22 | never-measured for occupancy | antibody + experiment | | **EXO1** | **no ENCODE target record** | never-measured | and ChIP is the wrong assay - resection is END-seq/BLESS/ssDNA-qPCR | | **DNA2** | **no ENCODE target record** | never-measured | same | **Every row's remaining cell now says the same thing, and that is the point.** Before this cycle the absence had a cheap corner and an expensive corner, and the cheap corner was doing a lot of rhetorical work - it is the sentence Q32 called "the single most useful sentence in this entry". **That corner is gone.** There is no longer any public dataset that can advance this question. It costs wet work now, and the entry should be quoted that way. **Two bounds that were NOT tested and must not be assumed.** (1) **Cell type.** This was an ENCODE-only sweep. IHEC, Blueprint and GEO were not queried for a BRCA1 or CBX3 dataset in a CD34+ progenitor or a KMT2A-rearranged line, so "no blood-cell data beyond K562 and GM12878" is an assumption here, not a measurement. (2) **Assay class.** ChIP-seq is a poor instrument for EXO1 and DNA2 even in principle - they are transient, sub-stoichiometric and act on resected ssDNA, and their absence from ENCODE is partly an absence of anyone believing ChIP would work. **One thing the compute added that the literature bound could not.** The element is not merely unoccupied by these proteins; **CTCF is demonstrably on it in all six cell types** (conservative-IDR peaks, signalValue 87.0 to 295.3). So the negative is a statement about the proteins, not about whether the anchor exists in these cells - which is the confound a literature-only absence could never exclude.
| script | no script of its own - this entry records what remains after analyses/q35-brca1-cbx3-encode-at-the-kmt2a-locus |
|---|---|
| status | run |
| run date | 2026-09-22 |
| run by | Q35-encode-reanalysis |
| absence class | never-measured |
| absence class note | UPGRADED from `measured-not-analysed` by the Q35 compute. The headline class is now the SEVEREST of the set, because the least severe corner has been discharged with a negative result. Read the per-protein table before quoting anything. |
| what was discharged this cycle | the `measured-not-analysed` class for BRCA1 and CBX3, and for CBX1 which was never in the original set. The compute route to this question no longer exists. |
| cost of closing what remains | wet work. A DSB induced within 2 kb of the element (Cas9 or an integrated AsiSI/FokI site) in a CD34+ progenitor or a B-lineage progenitor line, with ChIP for BRCA1, CBX3 and CtIP at 5, 15, 30 and 60 min - the design already written into FALSIFIER 3 of hyp-the-menin-arm-and-the-ctcf-arm-license-consecutive-steps-of-one-resection-reaction. |
| relationship to existing entries | This does not replace `absence-no-resection-factor-localised-at-the-kmt2a-locus`, which remains correct as a literature bound and whose 21 queries stand. It records what changed when that entry's own `measured-not-analysed` prescription was carried out, and it exists separately because the original entry's headline class and its 'single most useful sentence' are now both false and a reader quoting either would be wrong. |
absence class per protein
{
"BRCA1": "resting state MEASURED and negative (7 ENCODE experiments, incl. K562 and GM12878, 2026-09-22); damage state never-measured in every cell type",
"CBX3": "resting state MEASURED and negative (4 ENCODE experiments, 3 passing the dynamic-range gate); damage state never-measured",
"CBX1": "resting state MEASURED and negative (2 ENCODE experiments); damage state never-measured. Added by Q35; not in Q32's original six",
"KDM5A": "never-measured in a blood cell; measured-not-analysed in A549, H1 and HepG2. Unchanged by this cycle - Q35 did not run the KDM5A tracks",
"BARD1": "CORRECTED 2026-09-22. never-measured for OCCUPANCY, but the stated reason was false: /targets/BARD1-human/ returns HTTP 200 and the 404 came from the assay_title=TF+ChIP-seq filter on the search endpoint. BARD1 also has two released CRISPR RNA-seq experiments.",
"EXO1": "never-measured - no ENCODE target record, and ChIP-seq is arguably the wrong assay class",
"DNA2": "never-measured - no ENCODE target record, same assay-class caveat"
}what was NOT tested
[ "IHEC, Blueprint and GEO were not swept for a BRCA1 or CBX3 dataset in a CD34+ progenitor or a KMT2A-rearranged line. The cell-type bound in this entry is an ENCODE bound, not an archive bound.", "KDM5A's three ENCODE experiments were not run. They are in a lung, an embryonic stem and a liver cell, so a negative there would not answer the blood-cell question, but the run is cheap and was not done.", "CBX5/HP1-alpha's four ENCODE experiments were not re-run at signal level by Q35; Q110 covered two of them at peak level.", "No damage-state dataset of any kind was searched for, in any archive, for any of the seven proteins." ]
- Does any archive other than ENCODE hold a BRCA1 or CBX3 occupancy dataset in a CD34+ progenitor or a KMT2A-rearranged line? One IHEC and one ChIP-Atlas query would convert this entry's main caveat into a bound, and neither was run.
- Is there any deposited dataset, anywhere, in which a resection factor was ChIPped in a cell after DNA damage at a mapped locus? If END-seq and BLESS depositions are counted, the damage-state class may be `measured-not-searched` rather than `never-measured`, which is a difference of orders of magnitude in cost.
- Should KDM5A's three ENCODE experiments be run anyway? A negative in lung, ES and liver would not answer the blood question but would make the KDM5A row symmetric with the other three, and the compute is an afternoon.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
Across all 13 released ENCODE BRCA1, CBX3 and CBX1 ChIP-seq experiments there is no called peak in the KMT2A gene, and at signal level the breakpoint cluster core reads 0.57-1.24x the genome-wide median of same-width tiles in the 11 tracks passing a pre-registered dynamic-range gate.
ENCODE Portal: all 13 released human BRCA1, CBX3 and CBX1 TF ChIP-seq experiments, signal and peak files, downloaded and analysed 2026-09-22 · ENCODE Project Consortium · ENCODE Portal 2026What this evidence cannot show
Resting-state, undamaged, non-haematopoietic-progenitor cell lines only. This bounds the resting state in the cells ENCODE holds and says nothing about the damage state, which is the state the mechanism is about.
Evidence that argues against this (2)
- A resting-state negative is exactly what the damage-conditional model predicts, so this result cannot discriminate between 'the cascade does not operate here' and 'the cascade operates here only after a break'. The entry it constrains says so in its own mechanism description. ENCODE Project Consortium 2026
- SCOPE BOUNDARY - THE ID IS WIDER THAN THE CLAIM. The id says 'any resection factor'; the claim is scoped by its own fields to the named proteins (BRCA1, BARD1, EXO1, DNA2, CBX3/CBX1, KDM5A). RPA is NOT among them, and RPA binding at the 11q23 translocation breakpoint hotspot after etoposide HAS been measured: ATM deficiency raised RAD51 and also RPA and INO80 binding there (PMID 21048951, 2010; the abstract confirms the RPA binding, the assay format was read from full text by the Q192 agent and not re-verified here). Do not quote this entry's id as covering RPA or any other recruited resection factor. The id was NOT renamed: it has many inbound references and public URLs, and the title and fields are already scoped. PLoS One 2010
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
CORRECTED 2026-09-22. This claim previously read 'ENCODE holds no target record at all for BARD1, EXO1 or DNA2 - the experiment search endpoint returns HTTP 404 for each'. The 404s are real but the inference was wrong for BARD1. Measured live on both endpoints: /targets/BARD1-human/ 200, /targets/EXO1-human/ 404, /targets/DNA2-human/ 404. A 404 from the FILTERED experiment search describes the query; the bare target record is a separate call and is the one that speaks about the protein. So EXO1 and DNA2 are never-measured in ENCODE, and BARD1 has a target record plus two released CRISPR RNA-seq experiments and no TF ChIP-seq.
ENCODE Portal experiment search for target.label BARD1, EXO1 and DNA2, TF ChIP-seq, released, Homo sapiens - HTTP 404 for all three, 2026-09-22 · ENCODE Project Consortium · ENCODE Portal 2026What this evidence cannot show
A 404 from this endpoint is how the portal reports an empty result set, which is not obvious and is easy to mistake for a transport failure. The queries that returned counts and the queries that returned 404 were run in the same loop with the same retry policy, so the 404s are not a network artefact - but they were verified as empty results rather than as errors, and that distinction is the whole content of this evidence item.
Evidence that argues against this (1)
- ENCODE is not the only ChIP archive, and an absent ENCODE target record does not mean nobody has ChIPped the protein. GEO and ChIP-Atlas were not queried for BARD1, EXO1 or DNA2 in this beat, so this bound is ENCODE-only. ENCODE Project Consortium 2026
classified as the corpus's only entry whose headline rests primarily on a thresholded peak set - the proven mode-1 exposure
the id and headline generalise to 'any resection factor'; a damage-state RPA/RAD51/INO80 occupancy measurement at this locus is published. The SCOPED six-protein claim inside the entry is untouched.
records what remains of the absence after that compute was run
the headline absence_class measured-not-analysed and its 'two of the six can be answered this week' sentence are both now false; the 21-query literature bound is unaffected
never-measured: no ENCODE target record (search returns HTTP 404)
never-measured: no ENCODE target record, and ChIP is arguably the wrong assay class
never-measured: no ENCODE target record, same assay-class caveat
unchanged this cycle: 3 ENCODE experiments, none in a blood cell, not run here
the remaining question is FALSIFIER 3's damage-state design, which now has no compute substitute
no ENCODE target record; ChIP is the wrong assay class for a transient nuclease
never-measured at this locus in any cell type
no ENCODE target record, unlike BRCA1 which has seven
never-measured at this locus; two literature records, both in other diseases
closes the first of its four what_was_NOT_tested items: IHEC/Blueprint/GEO were not swept, and the sweep changes four of seven per-protein classes
GSE218590 is a damage-state CtIP and MRE11 occupancy dataset, so 'for any resection factor' does not hold; the seven-protein table survives