start / graph / ana-kmt2a-exon-11-element-is-a-top-0-05-percent-etoposide-dsb-site

AnalysisstrongQ107-break-machinery-deposition

Read from deposited files today: the KMT2A exon-11 CTCF/cohesin element is a top-0.05% etoposide break site on chromosome 11 in a human B-cell-precursor line and in primary CD34+ marrow - and is NOT one in HCT116, the only cell line in which cohesin was ever degraded

also called the exon-11 element break readout · Q107 percentile readout · premise (i) measured · the HCT116 mismatch

We read the public data files for the exact 147 bases of the KMT2A gene that this project has been arguing about, in ten different published experiments, without downloading anything large. In blood-precursor cells and in real human bone-marrow progenitors, that spot is among the most break-prone one-in-two-thousand stretches of the whole chromosome. In the one cell line where scientists actually removed the ring protein to test whether it causes the breaks, the spot barely breaks at all.

The same thing, for a scientist

Every deposited human track from PMID 31202577 (Canela 2019, END-seq), PMID 28735753 (Canela 2017, END-seq) and PMID 31202576 (Gothe 2019, sBLISS) was range-read over HTTPS with `pybigtools` and the signal over **canonical *KMT2A* exon 11, GRCh37 chr11:118,359,329-118,359,475** (= GRCh38 chr11:118,488,614-118,488,760, offset exactly 129,285 bp, verified by Ensembl `map/human/GRCh38/.../GRCh37`) was ranked against all 918,411 non-overlapping 147-bp bins on chr11 in the same file - a statistic that needs no depth normalisation [E1][E2][E3]. **Etoposide-treated arms: Nalm6 (human B-cell-precursor ALL) 99.948th percentile, 475 bins above it; K562 99.944th (517 bins); CD34+ primary marrow 99.658th and 99.796th (3,142 and 1,875 bins); TK6 99.610th; MCF7 99.883rd. Untreated/DMSO arms fall to 85.8-95.9th. HCT116 + etoposide reaches only the 94.004th percentile (55,071 bins above it) - BELOW the 92.9th percentile its own no-etoposide control occupies once depth is accounted for** [E1][E2]. In the same cells, HCT116 CTCF and RAD21 ChIP put the element in the **top 0.02% of chr11** (99.981st and 99.980th percentiles; 176 and 185 bins above), with summits at GRCh37 chr11:118,359,409 and 118,359,392 - **both inside canonical exon 11** [E3]. **Transcription dependence, measured at this element:** DRB pre-treatment moves TK6 from the 98.016th to the **70.995th** percentile [E1]. **Conclusion: premise (i) of the Q105 triad is testable and partly answered - the element is a genuine TOP2 break hotspot wherever *KMT2A* is a haematopoietic gene, but the one loss-of-cohesin experiment was performed in the one cell type where it is not a hotspot at all.**

Go deeper

The readout, one table, rank-based so no normalisation choice can move it

Region: canonical KMT2A exon 11, GRCh37 chr11:118,359,329-118,359,475 (147 bp). Null: all 918,411 non-overlapping 147-bp bins on GRCh37 chr11 (135,006,516 bp), from the same file, uncovered bases counted as 0.

study	GSM	cell	treatment	assay	exon-11 mean	chr11 percentile	bins >= it
Canela 2017	GSM2635574	Nalm6 (human B-cell precursor ALL)	etoposide	END-seq	0.5521	99.948	475
Gothe 2019	GSM3444989	K562	ETO 30 uM 4 h	sBLISS	12.354	99.944	517
Canela 2017	GSM2635568	MCF7	etoposide	END-seq	0.3927	99.883	1,079
Gothe 2019	GSM3687239	primary CD34+ marrow rep2	ETO	sBLISS	21.333	99.796	1,875
Gothe 2019	GSM3687237	primary CD34+ marrow rep1	ETO	sBLISS	16.415	99.658	3,142
Gothe 2019	GSM3444985	TK6	ETO 30 uM 4 h	sBLISS	12.306	99.610	3,582
Gothe 2019	GSM3444993	TK6 untreated-then-ETO rep1	ETO	sBLISS	4.191	98.016	18,219
Canela 2017	GSM2635575	Nalm6	none	END-seq	0.1050	95.948	37,217
Canela 2019	GSM3714903	HCT116 WT	ETO 50 uM 30 min	END-seq	0.1175	94.004	55,071
Canela 2019	GSM3714905	HCT116 WT	none	END-seq	0.2052	92.893	65,271
Gothe 2019	GSM3687236	primary CD34+ marrow rep1	DMSO	sBLISS	0.9048	85.828	130,158
Gothe 2019	GSM3444992	TK6 + DRB then ETO rep1	ETO	sBLISS	0.8844	70.995	266,382
Canela 2019	GSM3714904	HCT116 RAD21-AID + auxin	ETO 50 uM 30 min	END-seq	0.0213	26.941	670,982

Raw means are NOT comparable across studies. Gothe's bigWigs are UMI-DSB counts extended +/-100 bp; Canela's are bedtools genomecov coverage. The percentile column is the comparable one, and the 147-bp bin width was chosen to equal the region's width so the observed value and the null are the same statistic.

Depth-normalised enrichments in the same files, for readers who want an effect size (region mean / that file's genome-wide mean from its bigWig header): Nalm6 ETO 10.77x vs NT 1.81x; MCF7 ETO 7.14x vs NT 2.88x; CD34+ ETO 5.75x / 6.95x vs DMSO 0.82x / 2.91x; K562 ETO 6.92x vs DMSO 0.66x; TK6 ETO 5.68x / 2.42x vs DMSO 0.00x / 0.00x; TK6 untr-ETO 2.33x / 1.82x vs DRB-ETO 0.65x / 0.65x; HCT116 WT-ETO 1.74x vs NT 1.48x vs RAD21-AID 0.32x.

Where the signal sits, to the base pair - and it is the element, not the cut point

Positions of the maximum within the 1,048-bp ENCODE CTCF+RAD21+SMC3 cluster (GRCh37 chr11:118,358,874-118,359,921; GRCh38 118,488,159-118,489,206):

track	argmax GRCh37	argmax GRCh38	inside canonical exon 11?	inside the Atkin 300-bp window?
HCT116 RAD21 ChIP	118,359,392	118,488,677	yes	yes
HCT116 CTCF ChIP	118,359,409	118,488,694	yes	yes
Nalm6 END-seq + ETO	118,359,390	118,488,675	yes	yes
MCF7 END-seq + ETO	118,359,387	118,488,672	yes	yes
CD34+ sBLISS + ETO rep1	118,359,337	118,488,622	yes	yes
CD34+ sBLISS + ETO rep2	118,359,358	118,488,643	yes	yes
TK6 sBLISS + ETO rep1/rep2	118,359,267	118,488,552	no (intron 10, 62 bp upstream)	yes
K562 sBLISS + ETO	118,359,265	118,488,550	no (intron 10, 64 bp upstream)	yes

All eight maxima fall inside the 300-bp patient break-prone window of PMID 34405474 (GRCh38 chr11:118,488,500-118,488,800), and all eight are within 144 bp of each other. The occupancy summits and the break maxima in the two haematopoietic-lineage END-seq arms agree to within 2-22 bp.

And the functional cut point is not where the signal is. Over the cut point +/- 500 bp (GRCh37 chr11:118,355,076-118,356,077 = GRCh38 118,484,861 +/- 500) the depth-normalised enrichment is 0.06-2.18x across all thirteen tracks, against 1.82-10.77x at exon 11 in the etoposide arms - and in the HCT116 ChIP the cut-point window is 1.20x (CTCF) and 0.27x (RAD21) against 84.94x and 104.12x at exon 11. The weak ENCODE score-514 CTCF cluster around canonical exon 10 (GRCh38 chr11:118,485,122-118,486,000) reaches 0.00-1.45x in every break track 123.

This independently settles the one inferential link Q105 flagged. Q105 assigned Gray 2023's 'exon 11' to canonical exon 11 rather than DCAL exon 11 (= canonical exon 10) on ENCODE occupancy strength alone, and recorded that the assignment was settleable only from deposited coverage nobody had read. Two independent CTCF/cohesin ChIP libraries in a real human cell line now put summits at GRCh38 chr11:118,488,677 and 118,488,694 - inside canonical exon 11 - in the top 0.02% of chr11, with essentially nothing at canonical exon 10. The Q105 assignment holds, and it now rests on a measured summit rather than on a cross-cell-type cluster interval.

What this does to premise (i) of the Q105 triad

Premise (i) as Q105 wrote it: occupancy up implies DSBs up.

The correlational half is now strongly supported, in the right cell types. The element carries a top-0.02% CTCF/RAD21 peak and a top-0.05% etoposide DSB peak, and the break maximum sits on the occupancy summit. In a human B-cell-precursor leukaemia line (Nalm6) and in primary human CD34+ bone-marrow progenitors - the closest available proxies for the cell of origin - it is one of the few hundred most break-prone 147-bp windows on the whole chromosome.

The causal half is still untested, and the reason is a cell-type mismatch, not a missing dataset. The only loss-of-cohesin human experiment in this literature is Canela 2019's HCT116 RAD21-AID arm, and in HCT116 the element sits at the 94.0th percentile with etoposide and the 92.9th percentile without it. Applying the paper's own peak-calling rule - >10-fold over the matched non-treated sample - the element is not a peak in HCT116 at all: the etoposide/untreated ratio there is below 1 once library depth is taken into account (1.74x vs 1.48x genome mean). You cannot measure the effect of removing cohesin on a break that does not happen.

The apparent 5.5-fold drop from GSM3714903 to GSM3714904 (0.1175 to 0.0213 raw; 94.0th to 26.9th percentile) is a change between two values that are both background in that cell type, measured once each, in two different clones, with no untreated control for the depleted one. It is directionally consistent with premise (i) and it is not evidence for it.

The experiment that would test premise (i) is the same experiment in a haematopoietic cell. See exp-cohesin-degron-break-map-in-a-human-b-precursor-at-kmt2a.

The within-side contradiction recorded in mechanism-loop-anchor-dsb-hotspots, measured at this element

mechanism-loop-anchor-dsb-hotspots records that Canela 2017 calls loop-anchor breakage largely transcription-independent while Gothe 2019 says it scales with transcriptional output and requires elongation, and that the contradiction is unresolved. At this element specifically, the deposited data side with Gothe.

Gothe's DRB arm is deposited in full (TK6, DRB vs untreated, each with DMSO and etoposide, two replicates). Over exon 11: untreated-then-etoposide reaches the 98.016th percentile (rep1) and 1.82-2.33x genome mean (both reps); DRB-then-etoposide falls to the 70.995th percentile (rep1) and 0.65x genome mean (both reps) 1. Maximum UMI-DSB pileup in the 1,048-bp element: 5 and 4 without DRB, 2 and 1 with DRB.

That is a 27-percentile-point rank drop and a ~3x enrichment drop at a single named 147-bp element, in two replicates, from a deposited processed file. It is the strongest single-locus statement available on the within-side dispute, and it was sitting unread.

Caveat that matters: DRB inhibits CDK9 and therefore elongation globally, and Gothe's own aggregate plots show a genome-wide reduction in etoposide-induced DSBs on DRB, so a locus-specific claim needs a rank statistic rather than a raw one - which is why the percentile column is the one to read. The element's rank falls from the top 2% to the top 29%, so the drop is larger than the global shift, but this is n = 2 and no formal locus-versus-genome test was run.

Reproducing this - it is cheap and it was refused last cycle

Total data transferred: a few hundred kilobytes across 19 files totalling 2.9 GB on the server, because bigWig is an indexed format and pybigtools issues HTTP range requests against the GEO FTP/HTTPS endpoint. Nothing was downloaded in bulk and nothing was written to disk.

The tool is pybigtools 0.3.0 (Rust/BigTools, maturin wheels). It does not install under the system CPython 3.14 on this machine - no wheel exists for 3.14, so pip falls back to cargo rustc and the build fails. It installs in seconds under the uv-managed CPython 3.12.13 already present on the box. See ana-pybigtools-reopens-the-bigwig-route-refused-in-cycle-57, because a previous cycle recorded a costed refusal of exactly this compute on the grounds that pyBigWig would not build and no UCSC binaries were available. That refusal was correct about pyBigWig and wrong about the conclusion.

Facts and figures

script	pybigtools 0.3.0 under CPython 3.12.13, HTTP range reads against ftp.ncbi.nlm.nih.gov; scripts held in the beat scratchpad, not committed
status	run
result summary	Canonical KMT2A exon 11 is a top-0.05%-of-chr11 etoposide DSB site in Nalm6 (99.948th pct), K562 (99.944th), MCF7 (99.883rd), primary CD34+ marrow (99.796th/99.658th) and TK6 (99.610th), and only a 94.004th-percentile site in HCT116 with etoposide - below the 92.893rd percentile of its own untreated control once depth is accounted for. HCT116 CTCF and RAD21 occupancy at the same 147 bp is 99.981st/99.980th percentile with summits at GRCh37 chr11:118,359,409 and 118,359,392, both inside the exon. DRB moves TK6 from the 98.016th to the 70.995th percentile at this element.
null model	all 918,411 non-overlapping 147-bp bins on GRCh37 chr11 (135,006,516 bp) from the same bigWig, uncovered bases as 0; bin width set equal to the region width so observation and null are the same statistic

inputs

[
  "GSE121740/GSE121742 sBLISS bigWigs: GSM3444985, GSM3444989, GSM3444992, GSM3444993, GSM3444984, GSM3444986, GSM3444987, GSM3444988, GSM3444996, GSM3444997, GSM3687236, GSM3687237, GSM3687238, GSM3687239",
  "GSE99194 END-seq bigWigs: GSM2635568 (MCF7 ETO), GSM2635569 (MCF7 NT), GSM2635574 (Nalm6 ETO), GSM2635575 (Nalm6 NT)",
  "GSE129524 END-seq bigWigs: GSM3714903, GSM3714904, GSM3714905",
  "GSE129526 ChIP-seq bigWigs: GSM3714939 (RAD21), GSM3714940 (CTCF), GSM3714941 (input)"
]

region of interest

{
  "canonical_exon_11_grch38": "chr11:118,488,614-118,488,760",
  "canonical_exon_11_grch37": "chr11:118,359,329-118,359,475",
  "ctcf_rad21_smc3_cluster_grch38": "chr11:118,488,159-118,489,206",
  "ctcf_rad21_smc3_cluster_grch37": "chr11:118,358,874-118,359,921",
  "patient_break_window_grch38": "chr11:118,488,500-118,488,800",
  "patient_break_window_grch37": "chr11:118,359,215-118,359,515",
  "functional_cut_point_grch38": "chr11:118,484,861|118,484,862",
  "functional_cut_point_grch37": "chr11:118,355,576|118,355,577",
  "grch38_minus_grch37_offset_bp": 129285,
  "transcript": "ENST00000534358 (canonical in both GRCh37 and GRCh38)",
  "measured_ctcf_summit_grch38": "chr11:118,488,694",
  "measured_rad21_summit_grch38": "chr11:118,488,677"
}

How this could bite us back

This readout makes the element look like a good therapeutic target for reducing therapy-related MLL rearrangement, which is exactly the trap Q105 named: the element is a CTCF/cohesin loop anchor, and Gray 2023's result is that losing CTCF occupancy there INCREASES KMT2A rearrangement. Lowering breakage by removing the anchor could raise the rearrangement rate.
Every percentile here is measured after pharmacological etoposide. A programme aimed at infant KMT2A::AFF1, which arises in utero without etoposide, would be importing a drug-trapping phenotype as a model of spontaneous fragility.
CTCF occupancy at this element in a human haematopoietic cell does not exist in any of the three deposits - but it does exist in ENCODE, and this beat read it: GM12878 CTCF puts the summit at the same base as HCT116, 113-135x genome mean. The residual gap is narrower and sharper than 'no blood cell' - it is that no B-cell precursor, CD34-positive or fetal occupancy measurement exists anywhere for this element, and the index disease originates in fetal liver.

Still open

Is KMT2A transcribed at all in lovastatin-arrested G1 HCT116? If it is silent there, the HCT116 negative is fully explained by Gothe's transcription requirement and is not a cell-lineage fact - and no nascent-RNA or RNA-seq sample for HCT116 exists in either deposit, so this needs an external track (ENCODE HCT116) to settle.
Do the deposited per-DSB BED files of GSE121740 - which carry UMI counts at base resolution rather than the +/-100 bp extension - place the CD34+ break maximum on the CTCF motif itself, and on which strand? That is a ~150 MB download per etoposide replicate and would give motif-level positioning the bigWigs cannot.
Is the element's break rank in Nalm6 and CD34+ higher than the ranks of the other ~89 MLL-recombinome partner loci in the same files? Gothe reports partner genes are enriched at loop anchors but does not rank KMT2A against them at single-element resolution.
The element's percentile in untreated Nalm6 is already 95.9th and in CD34+ DMSO 85.8th. Is that residual signal endogenous TOP2 activity, or coverage bias? The matched input for HCT116 reads 2.67x genome mean at this exon, which argues part of it is mappability.

Every source behind this entry (4)

E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
In the deposited sBLISS bigWigs of PMID 31202576 (GEO GSE121740/GSE121742, hg19), the 147 bp of canonical KMT2A exon 11 at GRCh37 chr11:118,359,329-118,359,475 ranks at the 99.944th percentile of all 918,411 non-overlapping 147-bp bins on chr11 in etoposide-treated K562 (517 bins above it), the 99.796th and 99.658th percentiles in two replicates of etoposide-treated primary human CD34+ bone-marrow progenitors (1,875 and 3,142 bins), and the 99.610th percentile in etoposide-treated TK6, against 85.828th in CD34+ DMSO; and DRB pre-treatment (200 uM, 3 h) moves TK6 from the 98.016th percentile (untreated-then-etoposide) to the 70.995th percentile (DRB-then-etoposide), with depth-normalised enrichment falling from 2.33x and 1.82x to 0.65x and 0.65x across the two replicates.
GEO GSE121742 / GSE121740 (SuperSeries and BLISS-seq SubSeries for Gothe et al. 2019), deposited per-sample bigWig tracks GSM3444985, GSM3444989, GSM3444992, GSM3444993, GSM3687236, GSM3687237, GSM3687239, read over HTTP range requests with pybigtools 0.3.0 on 2026-09-22 · Gothe HJ, Bouwman BAM, Gusmao EG, Piccinno R, Sayols S, Drechsel O, Petrosino G, Minneker V, Josipovic N, Mizi A, Nielsen CF, Wagner EM, Takeda S, Sasanuma H, Hudson DF, Kindler T, Baranello L, Papantonis A, Crosetto N, Roukos V · NCBI GEO (data accompanying Mol Cell 75:267-283.e12) 2019
PMID 31202576 DOI
What this evidence cannot show
Signal values in these files are UMI-DSB counts extended +/-100 bp by the submitters, so a single break contributes 201 bp of signal and the effective resolution is ~200 bp, not 1 bp; the argmax positions are therefore accurate to roughly +/-100 bp and the TK6/K562 maxima reported 62-64 bp upstream of the exon boundary are within that resolution of the exon itself. Percentile is computed against chr11 only, not the genome, and chr11 includes a large pericentromeric artefact region (the top bins of every Canela END-seq track on chr11 cluster at chr11:51,579,507-51,593,031, present in untreated samples too), which makes the percentile conservative rather than generous. The DRB result is n = 2 with no formal locus-versus-genome-shift test; DRB reduces etoposide-induced DSBs globally in this dataset, so part of the rank drop is the global effect. No replicate exists for K562. CD34+ cells are pooled primary bone marrow from adult donors, not fetal liver, and not B-lineage-committed.
Evidence that argues against this (1)
- The strongest argument against reading these numbers as a KMT2A-specific fragility statement is that etoposide is a pharmacological trap, not a physiological exposure: every one of these percentiles is measured after 20-60 uM etoposide, and the untreated/DMSO arms put the element at only the 85.8th-95.9th percentile. So what is demonstrated is that the element is a preferred site of drug-trapped TOP2 cleavage, which is directly relevant to therapy-related AML but is an indirect proxy for the spontaneous fetal breakage that produces infant KMT2A::AFF1. Gothe's own text makes the same point in reverse - TDP2 deficiency 'did not lead to genomic instability in absence to ETO-treatment'. Gothe HJ et al. 2018
E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
In the deposited END-seq bigWigs of PMID 28735753 (GEO GSE99194, hg19) canonical KMT2A exon 11 ranks at the 99.948th percentile of 918,411 chr11 147-bp bins in etoposide-treated Nalm6 (475 bins above it, 10.77x the file's genome-wide mean) against the 95.948th percentile in untreated Nalm6 (1.81x), and at the 99.883rd percentile in etoposide-treated MCF7 (7.14x); whereas in the deposited END-seq bigWigs of PMID 31202577 (GEO GSE129524) the same interval reaches only the 94.004th percentile in etoposide-treated WT HCT116 (55,071 bins above it, 1.74x genome mean), the 92.893rd percentile in untreated HCT116 (1.48x) and the 26.941st percentile in RAD21-AID HCT116 after auxin plus etoposide (0.32x).
GEO GSE99194 [END-seq] (Canela et al. 2017) samples GSM2635568, GSM2635569, GSM2635574, GSM2635575 and GSE129524 [END-seq] (Canela et al. 2019) samples GSM3714903, GSM3714904, GSM3714905 - deposited bigWig tracks read over HTTP range requests with pybigtools 0.3.0 on 2026-09-22 · Canela A, Maman Y, Huang SN, Wutz G, Tang W, Zagnoli-Vieira G, Callen E, Wong N, Day A, Peters JM, Caldecott KW, Pommier Y, Nussenzweig A · NCBI GEO (data accompanying Cell 170:507-521.e18 and Mol Cell 75:252-266.e8) 2019
PMID 31202577 DOI
What this evidence cannot show
Every arm is n = 1; GSE99194 deposits one etoposide and one untreated library per human cell type and GSE129524 deposits one per HCT116 condition. HCT116 was G1-arrested with 20 uM lovastatin for 24 h and pulsed with etoposide for 30 min, while Nalm6 and MCF7 were treated without cell-cycle synchronisation - so the HCT116-versus-Nalm6 comparison confounds cell type with cell-cycle state and pulse length, and the HCT116 result is the weaker claim of the two. The HCT116 arms use only the ExoVII+ExoT END-seq variant, which reports all TOP2 lesions rather than DSBs alone. Percentiles are chr11-only and include the pericentromeric artefact, which is conservative. Nalm6 carries its own rearrangements and is a cultured leukaemia line, not a normal B-cell precursor.
Evidence that argues against this (1)
- The cleanest counter-reading is that the HCT116 result is not a cell-type fact but a protocol fact: 30 min of etoposide versus 4 h in Gothe, in lovastatin-arrested G1 cells. TOP2 cleavage-complex conversion to irreversible DSBs is time-dependent in Canela 2019's own time course (0.5 h to 12.5 h), so a 30-min pulse may simply be too short at this element. That reading is testable within the existing deposit - the mouse pre-B time course GSM3714871-GSM3714876 exists - and it would weaken, not overturn, the conclusion, because the comparison that matters (HCT116 WT+ETO versus HCT116 NT) is internal to the same protocol and still shows no induction. NCBI Gene Expression Omnibus 2019
E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
In the deposited HCT116 ChIP-seq bigWigs of GEO GSE129526, canonical KMT2A exon 11 (GRCh37 chr11:118,359,329-118,359,475) ranks at the 99.981st percentile of 918,411 chr11 147-bp bins for CTCF (176 bins above it; 84.94x the file's genome mean; 48.7-fold over the matched input) and the 99.980th percentile for RAD21 (185 bins; 104.12x genome mean; 70-fold over input), with summits at GRCh37 chr11:118,359,409 (CTCF) and chr11:118,359,392 (RAD21) - both inside the exon - while the same tracks read 1.20x (CTCF) and 0.27x (RAD21) over the functional cut point at GRCh37 chr11:118,355,076-118,356,077.
GEO GSE129526 [ChIP-seq] samples GSM3714939 (RAD21), GSM3714940 (CTCF) and GSM3714941 (input) in HCT116 - deposited bigWig tracks read over HTTP range requests with pybigtools 0.3.0 on 2026-09-22; interval conversion by Ensembl REST map/human/GRCh38/11:118488614-118488760/GRCh37 · Canela A, Maman Y, Huang SN, et al.; Nussenzweig A · NCBI GEO (data accompanying Mol Cell 75:252-266.e8) 2019
PMID 31202577 DOI
What this evidence cannot show
These samples are mis-annotated in GEO as Mus musculus/mm10; the files are human hg19 (chr1 = 249,250,621, chr11 = 135,006,516) - see ana-canela-2019-rad21-degron-end-seq-deposition-audit E2. All three ChIP samples list the same antibody catalogue number (Millipore 07-729, an anti-RAD21 reagent), so the CTCF track's reagent identity rests on the sample title alone; the two tracks are distinguishable as separate libraries by summit, maximum and genome mean. n = 1 per target, no replicate. HCT116 is a near-diploid MSI colorectal carcinoma line: the occupancy is real in HCT116 and is not a statement about a haematopoietic cell. The fold-over-input figures use the interval mean of GSM3714941 (0.117), which is itself 2.67x that file's genome mean at this exon - i.e. the exon has elevated mappability/coverage baseline, and the fold-over-input numbers already absorb that.
Evidence that argues against this (1)
- Occupancy in HCT116 does not license an occupancy claim in a haematopoietic cell, and NO study in this dispute measured one - GSE99195 (Canela 2017 ChIP-seq) is mouse and Mus spretus only with zero human samples, Gothe 2019 generated no ChIP at all, and Gray 2023's four CTCF tracks are in RPE. This beat therefore read ENCODE instead, and the check PASSES rather than failing: CTCF ChIP in GM12878, a human B-lymphoblastoid line, puts its maximum at the identical base (GRCh38 chr11:118,488,694, ENCFF734CUT) with the 147-bp exon at 113x that file's genome mean; a second GM12878 CTCF experiment reads 135x with its maximum 23 bp away, and GM12878 RAD21 reads 35x. So the occupancy is not a colorectal artefact and is if anything stronger in B lineage. See absence-no-human-ctcf-or-cohesin-occupancy-measurement-in-a-blood-cell-at-kmt2a E3. A B-cell PRECURSOR, a CD34-positive progenitor and a fetal measurement all remain absent. ENCODE Project Consortium; NCBI Gene Expression Omnibus 2020
E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
The GRCh38-to-GRCh37 offset across this locus is exactly 129,285 bp and is constant over all three intervals of interest: the ENCODE CTCF+RAD21+SMC3 cluster GRCh38 chr11:118,488,159-118,489,206 maps to GRCh37 chr11:118,358,874-118,359,921; canonical KMT2A exon 11 GRCh38 chr11:118,488,614-118,488,760 maps to GRCh37 chr11:118,359,329-118,359,475; and the functional cut point GRCh38 chr11:118,484,861|118,484,862 maps to GRCh37 chr11:118,355,576|118,355,577 - all on the plus strand with widths preserved, with KMT2A occupying GRCh37 chr11:118,307,205-118,397,539 and ENST00000534358 canonical in GRCh37 as well as GRCh38.
Ensembl REST assembly mapping endpoint map/human/GRCh38/11:.../GRCh37 for three intervals, plus GRCh37 lookup of ENSG00000118058 and ENST00000534358, retrieved 2026-09-22 · Ensembl REST API; Ensembl release GRCh37 archive · Ensembl 2026
What this evidence cannot show
For the 1,048-bp cluster and the 147-bp exon the endpoint returns TWO mappings with identical coordinates - one onto chromosome 11 and one onto the alternative scaffold HG122_PATCH. For the 2-bp cut point it returns only the chromosome-11 mapping. A consumer taking 'the first mapping' gets the right answer here by luck of ordering; a consumer counting mappings would conclude the interval maps ambiguously. Always filter on seq_region_name == '11'.
Evidence that argues against this (1)
- A constant offset across a 4 kb span is what one expects inside a single unbroken alignment block, so this does not license extrapolating the same 129,285 bp offset elsewhere on chr11 or in KMT2A - the offset changes at every assembly break. Every coordinate in this entry was mapped individually rather than arithmetically, and the arithmetic agreement was used only as a consistency check. Ensembl 2026

Connected to

← supportsCanela 2019's full text never names MLL or KMT2A, and its HCT116 cohesin arm is lovastatin-arrested in G1 - so the only directional cohesin experiment was run in a cell where transcription-coupled fragility should be lowest
explains the HCT116 outlier in that table (94.004th percentile against 99.6-99.9 in every blood-lineage cell) by a condition the paper states: G1 arrest under lovastatin
depends on →Canela 2019's RAD21-degron END-seq IS deposited and IS readable today - one sample per arm, no replicates, no minus-auxin control, hg19, bigWig coverage only
reads the files this audit enumerated
supports →Adjudicated: CTCF does not do opposite things at KMT2A - it does one thing, and this graph's contradiction is an untested premise about whether a break becomes a rearrangement
measured CTCF and RAD21 summits at GRCh38 chr11:118,488,694 and 118,488,677, both inside canonical exon 11, closing the adjudication's one inferential link
supports →Run in coordinates: the protective CTCF site and the patient break-prone window are ONE element, it is cohesin-bound, and the 'functional cut point' is 3.75 kb away - two of the graph's three exon-11s were the same exon and the third was not
supplies the CTCF peak summit that entry recorded as unobtainable, from HCT116 ChIP rather than from Gray's RPE tracks
supports →The CTCF site in KMT2A exon 11 - a conserved insulator element sitting exactly on the functional cut point of the breakpoint cluster region, whose occupancy decides whether the locus rearranges
top 0.02% of chr11 for CTCF and RAD21 occupancy and top 0.05% for etoposide DSBs in blood-lineage cells
supports →Chromatin loop anchors are DNA-break hotspots - and the two papers that established it disagree about why
at this element the deposited DRB arm sides with Gothe: TK6 falls from the 98.0th to the 71.0th chr11 percentile on elongation inhibition
related to →KMT2A major breakpoint cluster region (BCR1): intron 7 - exon 13, ~8.3 kb
the break maximum is 3.7 kb from the functional cut point, inside the CTCF element, not at the junction
related to →KMT2A / MLL1 — the gene that breaks
canonical exon 11 of ENST00000534358 is the break-hotspot element
supports →What maps a DNA break, and the fact that none of the modern instruments has ever been pointed at KMT2A in a blood progenitor
converts that entry's assay survey into measured per-assay signal at the element
depends on →Canela 2017 deposited NO cohesin-depletion experiment - and it did deposit human END-seq in a B-cell-precursor leukaemia line, which nobody here had looked for
reads the Nalm6 and MCF7 arms this audit surfaced
depends on →Gothe 2019's deposit is all-human, has base-resolution break lists, includes primary CD34+ marrow and a transcription-inhibition arm - and contains no CTCF or cohesin perturbation of any kind
reads the sBLISS tracks this audit enumerated, including the CD34+ and DRB arms
← depends onThe cohesin-degron break map was run in the one human cell line where the KMT2A element does not break - so premise (i) is not underpowered, it is unaddressed
rests on the percentile readout across all thirteen human tracks
← supportsThe bigWig refusal of cycle 57 is overturned: pybigtools installs on this box under Python 3.12, reads GEO tracks by HTTP range request without downloading them, and passes the control loci the refusal itself demanded
the tooling that made the readout possible, with two API traps and a padded chrom.sizes trap named