Oncofusion

start / graph / genome-kmt2a-exon-11-ctcf-summit-rpe-gray

Non-codingstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q108-h3k9me3-at-the-bcr

Gray 2023's own 'exon 11 CTCF site' in RPE cells, measured: centroid GRCh38 chr11:118,488,662-118,488,761, inside canonical exon 11 and 34 bp from the independently measured summit - the graph's CTCF adjudication is confirmed, not overturned

also called Gray's RPE CTCF summit · the fourth independent measurement of the KMT2A exon-11 anchor · chr11:118,488,694 confirmed in RPE

A 2023 paper said a protein called CTCF sits on 'exon 11' of the KMT2A gene and protects it from breaking, but never gave a coordinate - and 'exon 11' means two different stretches of DNA 3.8 kb apart depending on whose numbering you use. Reading the paper's own raw data settles it: the paper means the same stretch this project had already measured, not the other one.

The same thing, for a scientist

PMID 37788669 makes CTCF occupancy at the *KMT2A* **exon-11** site the terminal step of its rearrangement brake and gives **no coordinates**. Because *KMT2A* exon numbering runs in at least five frames, 'exon 11' could denote Ensembl canonical exon 11 (GRCh38 chr11:118,488,614-118,488,760) or DCAL exon 11 = canonical exon 10 (chr11:118,484,862-118,484,975) - **3.8 kb apart**, and the distinction decides whether the protective element and the DCAL functional cut point are one object or two. Gray's four unique CTCF ChIP bigWigs in GSE210480 were read directly [E1]. **The signal-weighted centroid over the 1,048 bp CTCF/RAD21/SMC3 element is GRCh38 chr11:118,488,727.7 and 118,488,662.3 in the two control tracks and 118,488,760.6 and 118,488,676.4 in the two siKDM3B tracks - all four inside canonical exon 11, and the two control values within 34 bp of the independently measured summit chr11:118,488,694** [E1]. The maximum 200 bp bin sits at chr11:118,488,486 or 118,488,686 (adjacent bins) and is **3,624-4,024 bp away** from the DCAL cut point in every track. **Gray's 'exon 11 CTCF site' is canonical exon 11.** `genome-kmt2a-exon-11-ctcf-summit-measured`'s standing counter-argument - that Gray might mean an element 3.8 kb away and that a published adjudication in this graph might therefore be wrong - is **refuted by Gray's own deposit**, and this is now a four-experiment, three-cell-type agreement (HCT116, GM12878 x2, RPE).

Go deeper
The four measurements, and the resolution that bounds them
trackarmmax 200 bp bin (GRCh38)signal-weighted centroid (GRCh38)offset from chr11:118,488,694
GSM6431023 = GSM7594700siCT118,488,686118,488,727.7+33.7 bp
GSM6431025 = GSM7594702siCT118,488,486118,488,662.3-31.7 bp
GSM6431019 = GSM7594696siKDM3B118,488,886118,488,760.6+66.6 bp
GSM6431021 = GSM7594698siKDM3B118,488,486118,488,676.4-17.6 bp

All four centroids lie inside canonical exon 11 (chr11:118,488,614-118,488,760) or within 1 bp of its telomeric edge. The distance to the DCAL functional cut point (chr11:118,484,861|862) is 3,801-3,899 bp by centroid.

Why the centroid and not the argmax. These CTCF tracks are 200 bp binned (measured by run-length encoding a 20 kb window, modal run 200 bp in all four files). An argmax over a 200 bp grid is a bin label quantised to that grid, which is why the four maxima appear at 118,488,486 / 686 / 886 - three adjacent bins - and why the pre-registered +/-150 bp tolerance in design.py was narrower than one bin and initially returned the wrong verdict. The signal-weighted centroid over the whole element uses every bin and is not quantised, and it is the statistic quoted here.

Honest precision. 200 bp bins, n = 2 per arm, and a peak whose half-maximum width is 600-1,400 bp. The element is localised to about +/-50 bp by the centroid and to about +/-100 bp by the bin grid. What this measurement can and cannot distinguish: it cannot resolve 50 bp, and it emphatically can resolve 3.8 kb. The adjudication turns only on the latter.

Four experiments, three cell types, one element - and what remains unmeasured
sourcecell typeCTCF maximum (GRCh38)
Canela 2019, GSE129526HCT116, colorectal carcinomachr11:118,488,694
ENCODE ENCFF734CUT / ENCSR000AKBGM12878, B-lymphoblastoidchr11:118,488,694
ENCODE, second GM12878 experimentGM12878chr11:118,488,671
Gray 2023, GSE210480, this entryRPE, retinal pigment epitheliumchr11:118,488,662-118,488,761 (centroid, 4 tracks)

A 100 bp spread across four experiments in three unrelated lineages. This is now one of the better-localised noncoding elements in this graph.

Still unmeasured, and the gap has not moved. (a) Motif orientation - a coverage track gives no motif call, no CTCF PWM scan has been run on the 550 bp footprint, and Canela 2017's whole convergent-orientation argument makes it load-bearing. It remains a desk job and it remains undone. (b) A B-cell precursor, a CD34-positive progenitor or any fetal cell - GM12878 is a mature transformed B-lymphoblastoid line, not a precursor, and RPE and HCT116 are epithelial. Every cell in which this element has been measured is either transformed or non-haematopoietic.

And one thing this measurement adds that the summit alone did not. In Gray's own tracks the element is a 4.7x-chr11-mean CTCF site (6.96x over canonical exon 11 alone) while the DCAL cut point 3.8 kb away reads 1.27x - i.e. nothing. That reproduces, in RPE and in ratio-track units, the same contrast Q107 measured in HCT116 raw coverage (50x versus 3x). Two cell types, two normalisations, same verdict: the cut point is not a CTCF anchor.

Facts and figures
elementCTCF/cohesin loop-anchor site inside Ensembl canonical KMT2A exon 11, as measured in the founding paper's own cell type
track bin size bp200
positional precision bp50
unique ctcf tracks in the series4
cell typehTERT-immortalised human retinal pigment epithelial (RPE) cells - epithelial, non-haematopoietic, no KMT2A rearrangement
cell stateasynchronous proliferating culture, 72 h after control or KDM3B siRNA transfection
localizationchromatin, chr11q23.3, inside the KMT2A gene body; a CTCF/cohesin-bound element
subnuclear localizationnot measured here. A CTCF/cohesin anchor implies a loop or TAD-boundary position, and this graph records the KMT2A locus in that role elsewhere, but no subnuclear imaging or Hi-C was read for this entry
epigenetic readoutCTCF ChIP-over-input coverage at 200 bp resolution; the matched H3K9me1/2/3 histone readout for the same element is in ana-q108-h3k9me3-at-the-kmt2a-bcr-measured
shown for fusionNot shown for any fusion. The element was measured in unrearranged RPE cells; no KMT2A::AFF1 or other fusion-bearing cell was assayed, and the element sits in the intact KMT2A gene before any rearrangement exists.
motif orientationSTILL NOT MEASURED - a coverage track gives no motif call and no PWM scan has been run. Unchanged by this entry.
adjudication settledGray 2023's 'exon 11 CTCF site' is Ensembl CANONICAL exon 11, not DCAL exon 11 (= canonical exon 10). The two literatures describe ONE element, and the DCAL functional cut point 3.8 kb away carries no CTCF in Gray's own data.
grch38 minus grch37 offset bp129285
ctcf centroid grch38 control tracks
[
  "chr11:118488727.7",
  "chr11:118488662.3"
]
ctcf centroid grch38 sikdm3b tracks
[
  "chr11:118488760.6",
  "chr11:118488676.4"
]
ctcf max bin grch38
[
  "chr11:118488486",
  "chr11:118488686",
  "chr11:118488886"
]
offset from measured summit bp
[
  -31.7,
  33.7
]
distance to dcal cut point bp
[
  3801,
  3899
]
ctcf fold over chr11 mean control
{
  "canonical_exon_11": 6.96,
  "ctcf_element_1048bp": 4.67,
  "ctcf_summit_pm250": 6.45,
  "dcal_cut_point_pm250": 1.27,
  "bcr_core_7466bp": 1.86,
  "20kb_flanks": [
    0.89,
    0.97
  ]
}
How this could bite us back
  • Confirming the adjudication makes the element look settled, and it is settled only as a coordinate. The direction of effect at this element is still contested inside this graph: Gray 2023 predicts that weakening CTCF here RAISES KMT2A rearrangement, while the loop-anchor fragility literature makes CTCF-bound anchors the preferred sites of breakage. An intervention aimed at this now-precise coordinate could move risk either way.
  • Every cell type in which this element has been measured is transformed, immortalised or epithelial. A precise coordinate from four such experiments will travel as 'the KMT2A CTCF site' and be applied to an infant's fetal progenitor, where it has never been measured. The cell type must travel with the coordinate.
Still open
  1. What is the CTCF motif orientation at this element? Four experiments now agree on its position to within 100 bp and not one of them has produced a motif call. A JASPAR or HOCOMOCO PWM scan of the 550 bp footprint sequence is a desk job and it is still undone, and Canela 2017's convergent-orientation argument makes it load-bearing.
  2. Is this element CTCF-occupied in a fetal B-cell precursor or a CD34-positive progenitor? All four measurements are in transformed or non-haematopoietic lines. This is the gap `absence-no-human-ctcf-or-cohesin-occupancy-measurement-in-a-blood-cell-at-kmt2a` names, and it is untouched by adding a fourth epithelial or lymphoblastoid measurement.
  3. Do the RPE CTCF tracks contain a second, weaker anchor anywhere in the 8.3 kb BCR? This beat read the element and the cut point and did not scan the whole BCR for secondary peaks at 200 bp resolution. If a second anchor exists, 'the exon-11 site' is an incomplete description of the locus's loop architecture.
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    In Gray 2023's four unique deposited CTCF ChIP-seq bigWigs from GSE210480 (RPE cells, hg19), the signal-weighted centroid of CTCF signal over the CTCF/RAD21/SMC3 element is GRCh38 chr11:118,488,727.7 and 118,488,662.3 in the two control tracks and 118,488,760.6 and 118,488,676.4 in the two siKDM3B tracks - all inside or within 1 bp of canonical exon 11 and 17-67 bp from the independently measured summit chr11:118,488,694 - while the maximum 200 bp bin is 3,624-4,024 bp from the DCAL functional cut point chr11:118,484,862 in every track; the tracks are 200 bp binned, measured by run-length encoding.

    GEO GSE210480 samples GSM6431019, GSM6431021, GSM6431023 and GSM6431025 (RPE CTCF ChIP-seq, deepTools ChIP-over-input ratio bigWigs, hg19) read at base resolution over HTTP range requests with pybigtools 0.3.0, 2026-09-22; 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) 2023
    PMID 37788669DOI
    What this evidence cannot show

    n = 2 per arm and the tracks are 200 bp binned, so the positional precision is about +/-50 bp by centroid and the 17-67 bp spread between tracks is not resolvable. The files are deepTools ChIP-over-input RATIO tracks, so fold values are compressed relative to raw coverage and are not comparable in magnitude with Canela's HCT116 figures - only the POSITION is being compared. Gray's paper gives no coordinate and no peak file, so this entry infers what the paper meant from the paper's own deposited signal; it is an inference about authorial intent supported by a measurement, not a statement the authors made. RPE is epithelial and not KMT2A-rearranged. The GRCh37-to-GRCh38 conversion uses the 129,285 bp offset re-derived per interval in lift.py.

    Evidence that argues against this (1)
    • This measurement shows where CTCF binds in RPE; it does not show that the authors were thinking of that element when they wrote 'exon 11'. A paper can name an exon from a RefSeq or legacy numbering while its ChIP-seq peak happens to sit elsewhere in the same gene, and the honest alternative reading is that the authors used a numbering frame in which their peak is exon 11 by coincidence. What defeats that alternative here is that the only other candidate - DCAL exon 11 = canonical exon 10 - carries 1.27x-chr11-mean CTCF in the authors' own tracks, i.e. no peak at all, so there is no element there for them to have meant. Gray ZH et al. 2023
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    The element measured here is the same object as the one already recorded in this graph from two other sources: Canela 2019's HCT116 CTCF summit at GRCh38 chr11:118,488,694, ENCODE GM12878 CTCF at the identical base, and the ENCODE CTCF+RAD21+SMC3 cluster chr11:118,488,159-118,489,206 which contains Ensembl canonical exon 11 chr11:118,488,614-118,488,760 and Atkin's 300 bp patient break-prone window chr11:118,488,500-118,488,800.

    What this evidence cannot show

    This item cites existing entries rather than re-reading ENCODE and Canela from source; the Canela figure WAS independently reproduced for position (not magnitude) by this beat's own reading of the same files in analyses/verify-bigwig-route-and-ctcf-anchor-hct116/run.py, and the ENCODE GM12878 figures were not re-derived here.

    Evidence that argues against this (1)
    • Agreement across four experiments strengthens the position but not the biology: all four are CTCF ChIP in a transformed or non-haematopoietic line, so a consistent summit across HCT116, GM12878 and RPE is evidence that the motif is strong and constitutive, not evidence that it is occupied in a fetal B-cell precursor where the index disease originates. The absence entry that names that gap is unaffected by this measurement. Atkin ND et al. 2021
Connected to