BRCA1 at the KMT2A exon-11 CTCF anchor is real, cell-type-restricted, and survives the summit-matched null Q35 asked for: 93.7th and 92.3rd percentile of same-cell CTCF anchors in HeLa-S3 and H1, rising to 95.2 and 95.7 under width-and-TSS matching - and 34.0 and 45.1 in the only two blood lines, where ENCODE's 300,000-peak pre-IDR lists are also empty
also called the H1/HeLa-S3 BRCA1 signal, resolved · BRCA1 is not at the KMT2A anchor in blood
A repair protein really does sit on one small spot inside the KMT2A gene, strongly enough to beat 92 to 96 of every hundred comparable spots in the same cells - but only in a cervical cancer line and an embryonic stem line. In the two blood-cell datasets, the only ones from the tissue where this leukaemia starts, it is below average and completely absent from even the most permissive peak lists.
The same thing, for a scientist
**The question this closes, and it was asked in writing.** `ana-q35-brca1-cbx3-and-cbx1-are-not-at-the-kmt2a-breakpoint-cluster-in-any-encode-cell` carried the open question: *"Is the exon-11 BRCA1 signal in H1 and HeLa-S3 (6.95x and 9.16x at 147 bp, absent from both blood lines) a CTCF-proximal artefact or a genuine cell-type-restricted event? Ranking that 147 bp window against 147 bp windows centred on CTCF summits, rather than against the element, would settle it in one."* **It was ranked that way. It is not an artefact of CTCF proximity.** [E1] | BRCA1 track | cell | lineage | 147 bp pct vs summit147 anchors | width+TSS-matched pct | 1,048 bp pct (Q35's geometry) | element in ENCODE's pre-IDR files | |---|---|---|---:|---:|---:|---| | ENCFF690UED | HeLa-S3 | cervical carcinoma | **93.69** | **95.17** | 66.85 | **both reps**, global IDR 0.0588 | | ENCFF620MRE | H1 | embryonic stem | **92.28** | **95.66** | 70.83 | **both reps**, global IDR 0.577 | | ENCFF133FPU | HepG2 | hepatocellular | 70.61 | 75.20 | 47.27 | none (108,511-peak list) | | ENCFF191NNV | HepG2 | hepatocellular | 55.08 | 64.61 | 22.86 | **both reps**, global IDR 0.482 | | ENCFF493XJG | K562 | **erythroleukaemia** | 45.11 | 36.58 | 64.57 | **none, 300,000-peak list** | | ENCFF017KXT | GM12878 | **lymphoblastoid** | 34.01 | 26.23 | 8.44 | **none, 300,000-peak list** | **Three things make this more than a rank.** **1. It is not CTCF-proximity.** In H1 the *CTCF* track itself places this anchor at only the **48.44th** percentile of its own anchors, while BRCA1 places it at the **92.28th**. BRCA1 is more enriched at this anchor than CTCF is. In HeLa-S3 the comparison is 78.67 (CTCF) against 93.69 (BRCA1). A CTCF-proximity artefact would track CTCF; this runs ahead of it [E1]. **2. Matching makes it sharper, not weaker.** Restricting the null to anchors within +/-25% of this anchor's peak width and within a factor of two of its distance to the nearest live-fetched Ensembl GRCh38 TSS moves H1 from 92.28 to **95.66** (n = 10,596 matched anchors) and HeLa-S3 from 93.69 to **95.17** (n = 3,995). The usual fate of an apparent enrichment in this project is to dissolve under matching; this one does the opposite [E1]. **3. An independent pipeline reproduces the HeLa number.** ChIP-Atlas's uniform MACS2 re-processing of the same HeLa BRCA1 SRA experiment (SRX150631, a read-depth coverage track with no input normalisation, so a different statistic on different arithmetic) puts the element at the **92.23rd** percentile of the same HeLa-S3 CTCF anchor census - 1.5 points from ENCODE's fold-change answer [E2]. **And the part that matters most for this laboratory is the negative.** The two ENCODE blood lines are the two lowest at signal level (34.01 and 45.11, both below the median anchor) **and** the two with a clean zero against a 300,000-peak pre-IDR list, the most permissive product ENCODE publishes. K562 is CML blast crisis and GM12878 is an EBV-transformed lymphoblastoid line; **neither is a KMT2A-rearranged cell, neither is a progenitor, and there is no third blood line to ask.** So the cleanest reading of the whole dataset is: *BRCA1 occupies this anchor in some human cell types and does not occupy it in the two blood cell types that have been measured* - which is the least convenient possible answer for a model built on BRCA1 licensing repair at a leukaemia breakpoint, and it is stated here in that form rather than as the 92nd-percentile headline [E1].
| element | CTCF loop anchor inside Ensembl canonical KMT2A exon 11 |
|---|---|
| interval scored | chr11:118,488,614-118,488,760 (GRCh38, 147 bp) |
| null | 147 bp windows centred on the point-source summit of every conservative-IDR CTCF narrowPeak of the same cell type (42,947-54,654 anchors per cell; a census, not a sample) |
| statistic | mean of the ENCODE fold-change-over-control bigWig, bins=1, fillna=0 explicit |
| why this is not a ctcf proximity artefact | in H1 the CTCF fold-change track places this anchor at the 48.44th percentile of its own anchors while BRCA1 places it at the 92.28th; an artefact of CTCF proximity cannot exceed CTCF |
| window geometry dependence | at Q35's 1,048 bp peak-centre geometry the same two tracks read 70.83 and 66.85; the +21.5 and +22.9 point shift is the geometry, and Q35 asked for the summit geometry by name before this run |
per track — the numbers behind the chart
{
"ENCFF690UED": {
"cell": "HeLa-S3",
"lineage": "cervical adenocarcinoma",
"percentile": 93.69,
"matched_percentile": 95.17,
"n_matched": 3995,
"fold_over_anchor_median": 2.754,
"mean": 4.35,
"anchor_median": 1.58,
"elem1048_percentile": 66.85,
"ctcf_track_percentile_same_cell": 78.67
},
"ENCFF620MRE": {
"cell": "H1",
"lineage": "embryonic stem cell",
"percentile": 92.28,
"matched_percentile": 95.66,
"n_matched": 10596,
"fold_over_anchor_median": 2.146,
"mean": 3.755,
"anchor_median": 1.75,
"elem1048_percentile": 70.83,
"ctcf_track_percentile_same_cell": 48.44
},
"ENCFF133FPU": {
"cell": "HepG2",
"lineage": "hepatocellular carcinoma",
"percentile": 70.61,
"matched_percentile": 75.2,
"fold_over_anchor_median": 1.325,
"elem1048_percentile": 47.27,
"ctcf_track_percentile_same_cell": 79.98
},
"ENCFF191NNV": {
"cell": "HepG2",
"lineage": "hepatocellular carcinoma",
"percentile": 55.08,
"matched_percentile": 64.61,
"fold_over_anchor_median": 1.089,
"elem1048_percentile": 22.86,
"ctcf_track_percentile_same_cell": 79.98
},
"ENCFF493XJG": {
"cell": "K562",
"lineage": "chronic myelogenous leukaemia, blast crisis",
"percentile": 45.11,
"matched_percentile": 36.58,
"fold_over_anchor_median": 0.921,
"elem1048_percentile": 64.57,
"ctcf_track_percentile_same_cell": 62.43
},
"ENCFF017KXT": {
"cell": "GM12878",
"lineage": "EBV-transformed lymphoblastoid",
"percentile": 34.01,
"matched_percentile": 26.23,
"fold_over_anchor_median": 0.754,
"elem1048_percentile": 8.44,
"ctcf_track_percentile_same_cell": 28.15
}
}excluded by C1
{
"ENCFF431CIQ": {
"cell": "K562",
"C1_ratio": 4,
"percentile_if_included": 14.04
}
}independent pipeline check
{
"SRX150631": {
"archive": "ChIP-Atlas",
"cell": "HeLa",
"target": "BRCA1",
"track_type": "read-depth coverage, no input control",
"percentile_vs_same_HeLa-S3_CTCF_anchor_census": 92.23,
"fold_over_anchor_median": 2.63
}
}blood line absence
{
"n_blood_cell_types_with_any_BRCA1_track": 2,
"cell_types": [
"K562 (CML blast crisis)",
"GM12878 (EBV-transformed lymphoblastoid)"
],
"signal_percentiles": [
45.11,
34.01
],
"peak_level": "zero at chr11:118,488,614-118,488,760 in every released GRCh38 peak file of ENCSR000DZS, ENCSR223MLH and ENCSR946BXO, including five 300,000-peak 'peaks and background as input for IDR' files",
"cells_that_do_not_exist_in_any_archive": "no BRCA1 ChIP-seq in a KMT2A-rearranged line, a CD34+ progenitor or a fetal haematopoietic cell, in ENCODE, ChIP-Atlas, ReMap or GEO"
}- If this is read as support for combining a repair-pathway inhibitor with a menin inhibitor in KMT2A-rearranged leukaemia, note that the occupancy it rests on was measured in a cervical carcinoma line and an embryonic stem line, and is ABSENT in both blood lines available. The strongest version of this finding is evidence against the blood-cell relevance of the model, not for it.
- BRCA1 is a tumour suppressor and a core homologous-recombination factor. Any intervention that removed BRCA1 from this anchor to prevent a translocation would remove it genome-wide, and the predictable consequence is the BRCA1-null phenotype - genome instability and, in the therapy-related setting, more rearrangements rather than fewer.
- The one clinical signal this graph already carries in this direction runs the wrong way: BRCA1 carriers given a TOP2 poison have an elevated therapy-related leukaemia risk, which argues that less BRCA1 at a breakpoint means more rearrangement, not less.
- Is the blood-line absence a property of blood or a property of these two particular lines? K562 is CML blast crisis with a highly rearranged genome and GM12878 is EBV-transformed; neither is a normal or a KMT2A-rearranged haematopoietic cell, and a CUT&RUN for BRCA1 in SEM or RS4;11 remains the only way to ask the question in the disease cell type.
- Does BRCA1 occupancy at this anchor track the cell's replication state rather than its lineage? H1 and HeLa-S3 are the two fastest-cycling lines in the panel and BRCA1 is S-phase-enriched; an S-phase fraction covariate across the six tracks would test the simplest alternative explanation and was not available in this run.
- Would the same summit-matched geometry find BRCA1 above the 90th percentile at CTCF anchors generally in H1 and HeLa-S3, rather than at this one? The class base rate is the missing number - Q37 measured it at peak level (10.3% of HeLa-S3 CTCF anchors carry a BRCA1 peak) but nobody has measured the fraction of CTCF anchors above the 90th signal percentile, which is 10% by construction and therefore needs a different framing to be informative.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
Against a census of every conservative-IDR CTCF anchor of the same cell type at 147 bp summit-centred geometry, BRCA1 occupies the KMT2A exon-11 anchor at the 93.69th percentile in HeLa-S3 and the 92.28th in H1 - rising to 95.17 and 95.66 under width-and-TSS matching, and exceeding the CTCF track's own percentile for the same anchor in both cells - while reading 45.11 and 34.01 in the only two blood lines, where ENCODE's 300,000-peak pre-IDR lists are also empty at the element.
ENCODE BRCA1 ChIP-seq fold-change-over-control bigWigs ENCFF690UED (ENCSR000EDB, HeLa-S3), ENCFF620MRE (ENCSR000EBX, H1), ENCFF133FPU (ENCSR343RJR, HepG2), ENCFF191NNV (ENCSR000EDY, HepG2), ENCFF493XJG (ENCSR223MLH, K562), ENCFF017KXT (ENCSR000DZS, GM12878), read 2026-09-22 against CTCF anchor sets ENCFF533TKY, ENCFF497ZGV, ENCFF695QSJ, ENCFF041ODC and ENCFF796WRU · ENCODE Project Consortium · ENCODE Portal 2026What this evidence cannot show
Six tracks in four cell types, one track per cell except HepG2 and K562. The two above the 90th percentile are single tracks in single cell types with no within-cell-type replication available, and H1's global IDR of 0.577 means its replicate ranks are close to independent - this is a signal-level statement, and the peak-level evidence for H1 is much weaker than for HeLa-S3. BRCA1 ChIP-seq is a notoriously difficult experiment and antibody quality varies across these labs and years; no antibody characterisation was checked in this run. A percentile among CTCF anchors says nothing about absolute occupancy or about how many molecules are there.
Evidence that argues against this (1)
- Q35 measured the same two tracks against the same conservative-IDR CTCF peak sets and reported the element BELOW the anchor median in 10 of 13 tracks, at 0.34-1.25x, with H1 BRCA1 at 1.20x. Both runs are right: the disagreement is entirely the window geometry, 1,048 bp centred on the peak centre versus 147 bp centred on the summit. A result that depends this heavily on a windowing choice is not robust, and the only reason the narrow geometry is not post-hoc is that the corpus asked for it in writing in a cycle-63 entry. Q35-encode-reanalysis 2026
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
An independent uniform MACS2 re-processing of the same HeLa BRCA1 SRA experiment, served by ChIP-Atlas as a read-depth coverage bigWig with no input normalisation, places the element at the 92.23rd percentile of the same HeLa-S3 CTCF anchor census - within 1.5 percentile points of ENCODE's fold-change answer.
ChIP-Atlas 2025 update: 10-year anniversary of a data-mining platform for exploring epigenomic landscape. · Zou Z, Ohta T, Kasukawa T, et al. · Nucleic Acids Research 2026What this evidence cannot show
This is the same underlying reads, not an independent experiment, so it tests the processing and not the biology - two pipelines agreeing on one SRA deposit is a weaker statement than two experiments agreeing. The ChIP-Atlas track is read-depth coverage with no input control, so its percentile is inflated at accessible chromatin generally; it is quoted here only as a concordance check and is never pooled with the ENCODE fold-change numbers.
Evidence that argues against this (1)
- Cross-pipeline agreement on a single deposit can be manufactured by a shared artefact - a copy-number gain, a mappability feature or a hyper-ChIPable region would produce the same signal in both. The auxin-degron CTCF control in Q37 (SRX11361143 calls the element, SRX11361144 with CTCF acutely degraded does not) is the evidence against a hyper-ChIPable artefact, and it was run for CTCF and not for BRCA1. Q37-archive-sweep 2026
the per-experiment table this cell-type restriction is read from
extends Q37's HeLa-S3 BRCA1 observation to H1 and HepG2, and adds the negative that matters for a leukaemia claim: no hit in any of 13 files in GM12878 or of 13 and 8 in the two K562 BRCA1 experiments
BRCA1 occupancy at the KMT2A exon-11 CTCF anchor, present in three non-haematopoietic cell types and absent from both blood lines