Oncofusion

start / graph / genome-kmt2a-exon-11-ctcf-anchor-ranked-among-matched-rpe-anchors

Non-codingstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q111-ctcf-anchor-ranking

The KMT2A exon-11 CTCF anchor, placed in its population: baseline strength 4.92x (91.7th percentile of 12,443 RPE anchors), KDM3B-depletion ratio 0.396x, rank 58 of 1,625 strength-matched anchors - a strong, strongly-responding anchor at the edge of a continuum, not an outlier

also called the anchor's percentile, both axes · 91.7th for strength, 3.6th for loss · rank 58 of 1,625

This is the record of where the loop-anchor inside the KMT2A gene sits among all twelve thousand similar sites in the same cells - how strong it is to begin with, and how much protein it loses when the enzyme is removed. It is stronger than 92 out of 100 of them and loses more than 96 out of 100 of the ones its own strength, which puts it at the edge of the crowd rather than outside it.

The same thing, for a scientist

Three earlier entries locate this element: `genome-kmt2a-exon-11-ctcf-site` records it as an ENCODE cluster interval, `genome-kmt2a-exon-11-ctcf-summit-measured` places the CTCF summit at GRCh38 chr11:118,488,694 and the RAD21 summit at 118,488,677 from HCT116 data, and `genome-kmt2a-exon-11-ctcf-summit-rpe-gray` confirms the position in Gray's own RPE tracks by signal-weighted centroid. **What none of them records is the element's place in a population**, and every specificity claim made about it needs one [E1]. **Position, in a fourth independent readout.** Calling anchors genome-wide on the 200 bp grid of the same four RPE CTCF bigWigs, the anchor's argmax bin is hg19 chr11:118,359,400 = **GRCh38 chr11:118,488,685** - **9 bp** from the independently measured HCT116 summit and inside its own 200 bp bin. That is now four experiments and three cell types (HCT116, GM12878 x2, RPE) agreeing within ~100 bp, plus this argmax [E1]. **Strength.** Baseline CTCF over a fixed 1,000 bp window centred on that bin is **4.92x** the track's own genome mean, the **91.7th percentile** of 12,443 anchors, whose range is 2.51x to 14.06x. **It is in the top decile and it is not near the top of it** - the strongest anchor in RPE is nearly three times stronger [E1]. Note that this is a *ratio-track* magnitude and is not comparable with the 50.05x CTCF and 72.22x RAD21 figures recorded for this element from raw HCT116 coverage in `ana-kmt2a-exon-11-ctcf-anchor-is-one-cohesin-bound-element`. **Response.** siKDM3B/control over the same window is **0.396x**, with both siKDM3B replicates below both controls. Rank **58 of 1,625** anchors matched within +/-10% of its strength (3.57th percentile; matched median 0.633), **258 of 12,443** overall (2.07th percentile; median 0.696) [E1]. **Precision, which is the limiting fact.** The two control replicates disagree two-fold at this element - 9.24x and 4.69x of chr11 mean at canonical exon 11 - so the four per-replicate ratios span **0.200x to 0.499x**, and the pre-declared split-replicate control places the anchor at the **31.9th** percentile rather than the 3.6th [E2]. The direction is unambiguous in every pairing; the magnitude is not resolved at n = 2 versus n = 2. **The 90 kb gene span is flat (0.956x), so whatever happens is local to the anchor** [E2].

Go deeper
The two axes, and why both are needed
axisvaluepercentile among 12,443 RPE anchors
baseline CTCF strength, 1,000 bp window4.92x91.7%
siKDM3B/control ratio0.396x2.07% (most-losing end)

A claim about this element can fail on either axis. Reporting only the strength makes it a strong anchor among many; reporting only the ratio makes it a big responder among many; reporting both, matched, is what makes it the strong end of a continuum - and the matched median of 0.633 is the number that decides the argument, because it says the typical anchor of this strength also loses a third of its CTCF.

Where the element is NOT exceptional:

  • It is not the strongest anchor in RPE (14.06x is).
  • It is not the most CTCF-losing anchor (0.21x at hg19 chr4:84,566,000 is).
  • It is not even the most CTCF-losing anchor on chr11: hg19 chr11:8,241,000 = GRCh38 chr11:8,370,285 has strength 4.95x - essentially identical to KMT2A's 4.92x - and a ratio of 0.281x, larger than KMT2A's loss, with both replicates consistent. A near-perfect strength match on the same chromosome that responds more, and nothing associates it with a fragile site 1.

Where it IS unusual, and this is the part that survives: a strong CTCF/cohesin anchor lying inside a 147 bp coding exon of a large gene, 3.8 kb from a functional cut point, inside a 300 bp patient break-prone window, with the whole 90 kb gene body flat while the anchor moves. The combination is what makes the locus interesting, and none of it is a property of the CTCF response magnitude.

What is still not measured at this element after four cycles of work on it

The motif has still never been scanned. genome-kmt2a-exon-11-ctcf-summit-measured and genome-kmt2a-exon-11-ctcf-summit-rpe-gray both name it as an undone desk job, and it remains undone: no PWM scan, no orientation call, no degeneracy score, despite Canela's convergent-orientation argument making orientation load-bearing for any loop-anchor claim. This beat did not do it either, and it is now the single most conspicuous gap in the element's dossier.

Also absent, and all of them are the material properties that hyp-the-kmt2a-anchor-is-hypersensitive-not-uniquely-targeted proposes as the basis for hypersensitivity:

propertystatus
CTCF motif score and orientationnever measured
CpG methylation state of the motif, any cell typenever read
replication timingnever read
cohesin load relative to matched anchorsmeasured in HCT116 as an absolute (72.22x) but never as a percentile
CTCF or cohesin occupancy in ANY haematopoietic cellabsent - see absence-no-human-ctcf-or-cohesin-occupancy-measurement-in-a-blood-cell-at-kmt2a
absolute occupancy rather than fold-over-controlunavailable from this deposit, whose Input samples have no file

And the comparison that would settle the hypersensitivity question is now cheap: the 1,624 strength-matched anchors are persisted in anchors.tsv with coordinates, so asking "is the KMT2A anchor unusual in motif score, methylation or replication timing among anchors that lose CTCF as much as it does?" is a join against public tracks, not a new experiment.

Facts and figures
elementCTCF/cohesin loop-anchor site inside Ensembl canonical KMT2A exon 11 (ENST00000534358), placed in its genome-wide population in the founding paper's own cell type
grch38chr11:118,488,159-118,489,206 (the 1,048 bp CTCF+RAD21+SMC3 element)
grch37chr11:118,358,874-118,359,921
grch38 minus grch37 offset bp129285
offset re derived this cycleyes - Ensembl REST, six intervals of width 2, 147, 501, 501, 1,048 and 90,341 bp, mappings filtered on seq_region_name == '11', offset constant at exactly 129,285 in all six
argmax bin grch38118488685
bp from the measured hct116 summit9
positional precision bp100
track bin size bp200
baseline ctcf strength fold over track genome mean4.916
baseline strength percentile among 12443 rpe anchors0.917
sikdm3b over control ratio0.396
gene span 90kb ratio0.956
what is unusual about this element and survivesa strong CTCF/cohesin anchor inside a 147 bp coding exon of a 90 kb gene, 3.8 kb from a functional cut point, inside a 300 bp patient break-prone window, with the whole gene body flat while the anchor moves - none of which is a property of its CTCF response magnitude
motif orientationSTILL NOT MEASURED after four cycles. No PWM scan, no orientation call, no degeneracy score. Named as an undone desk job by two earlier entries and not done by this beat either.
localizationchromatin, chr11q23.3, inside the KMT2A gene body; a CTCF/cohesin-bound element
subnuclear localizationnot measured. A CTCF/cohesin anchor implies a loop or TAD-boundary position and this graph records the KMT2A locus in that role elsewhere, but no imaging or Hi-C was read for this entry and no subnuclear position is claimed
cell typehTERT-immortalised human retinal pigment epithelial (RPE) cells - epithelial, non-haematopoietic, no KMT2A rearrangement. Every percentile here is an RPE percentile.
cell stateasynchronous proliferating culture 72 h after control or KDM3B siRNA transfection
epigenetic readoutCTCF ChIP-over-control occupancy at 200 bp, as both a baseline-strength percentile and a perturbation-response percentile against 12,443 anchors called from the same control tracks
shown for fusionNot shown for any fusion. The element was measured in unrearranged RPE; it sits in the intact KMT2A gene before any rearrangement exists, and no fusion-bearing cell was assayed.
ratio rank and percentile
pre_declared_split_replic…185.0pre_declared_split_replicate_control: 185nearest_500_in_S500.0nearest_500_in_S: 500strength_decile_101,245strength_decile_10: 1245matched_S_within_10pct1,625matched_S_within_10pct: 1625all_anchors12,443all_anchors: 12443
anchor strength range in rpe
[
  2.509,
  14.056
]
ratio rank and percentile — the numbers behind the chart
{
  "matched_S_within_10pct": {
    "n": 1625,
    "rank": 58,
    "percentile": 0.0357,
    "matched_median": 0.633
  },
  "nearest_500_in_S": {
    "n": 500,
    "rank": 22,
    "percentile": 0.044,
    "matched_median": 0.622
  },
  "strength_decile_10": {
    "n": 1245,
    "rank": 61,
    "percentile": 0.049,
    "matched_median": 0.615
  },
  "all_anchors": {
    "n": 12443,
    "rank": 258,
    "percentile": 0.0207,
    "median": 0.696
  },
  "pre_declared_split_replicate_control": {
    "n": 185,
    "rank": 59,
    "percentile": 0.319,
    "ratio": 0.586,
    "matched_median": 0.685
  }
}
per replicate control spread at exon11
{
  "GSM6431023": 9.243,
  "GSM6431025": 4.688
}
per replicate sikdm3b at exon11
{
  "GSM6431019": 1.85,
  "GSM6431021": 2.339
}
four pairwise ratios at exon11
[
  0.2,
  0.253,
  0.395,
  0.499
]
closest chr11 strength match that responds more
{
  "hg19": "chr11:8,241,000",
  "grch38": "chr11:8,370,285",
  "strength": 4.954,
  "ratio": 0.281,
  "both_reps_consistent": true,
  "note": "a near-identical baseline strength on the same chromosome with a LARGER CTCF loss, and no association with any fragile site"
}
king third measurement of the ensembl mapping set
{
  "why_this_is_here": "this entry reports the Ensembl patch-mapping set as inverted against registry row 168 and correctly flagged it as a single unrepeated live query. The king re-measured, and got a third answer.",
  "the_kings_measurement_2026_09_22": {
    "2 bp DCAL cut point (width 2)": "1 mapping, chromosome 11 only",
    "canonical exon 11 (width 147)": "2 mappings, chromosome 11 + HG122_PATCH",
    "1,048 bp element (width 1048)": "2 mappings, chromosome 11 + HG122_PATCH",
    "CTCF summit +/-250 (width 501)": "2 mappings, chromosome 11 + HG122_PATCH",
    "BCR core (width 7466)": "2 mappings, chromosome 11 + HG122_PATCH",
    "gene span (width 90341)": "2 mappings, chromosome 11 + HG122_PATCH",
    "offset": "exactly 129,285 bp, CONSTANT across all six intervals"
  },
  "it_reproduces_Q107_not_this_beat": "element 2, exon 2, cut point 1 - Q107's pattern exactly, and the opposite of this beat's element 1 / cut point 2",
  "therefore": "the registry row was NOT dropped. Three observations exist, two agree and one differs, two of them from the same day - which supports this entry's own conclusion that the mapping set is not stable, rather than either party being wrong. The count of mappings is not a reproducible observable; the chromosome-11 mapping and the 129,285 bp offset are, and both are present in all three.",
  "nothing_measured_in_this_beat_is_affected": "this beat filtered on seq_region_name == '11' throughout and re-derived the offset across six widths, so its coordinates are correct under every observed pattern"
}
KING CORRECTION - the GRCh38 lift of the comparator anchor
{
  "THE_STATED_GRCh38_COORDINATE_IS_WRONG": "This beat states 'hg19 chr11:8,241,000 = GRCh38 chr11:8,370,285'. That is wrong. 8,370,285 - 8,241,000 = 129,285, which is the GRCh37->GRCh38 offset at the KMT2A locus at 118.4 Mb; it was applied here as though the offset were constant along chromosome 11, and it is not.",
  "the_correct_lift_measured_by_the_king_2026_09_22": {
    "hg19 chr11:8,241,000": "GRCh38 chr11:8,219,453, offset -21,547",
    "controls proving the offset varies": {
      "hg19 chr11:8,000,000": "GRCh38 7,978,453, offset -21,547",
      "hg19 chr11:50,000,000": "GRCh38 49,978,464, offset -21,536",
      "hg19 chr11:118,358,874 (the KMT2A element)": "GRCh38 118,488,159, offset +129,285"
    },
    "so": "the offset is locally constant but varies by more than 150 kb along chromosome 11 and changes sign; a single locus's offset is never transferable to another locus",
    "source": "Ensembl REST map/human/GRCh37/11:start..end:1/GRCh38, filtered on seq_region_name == '11'"
  },
  "NOTHING_MEASURED_IS_AFFECTED": "The GSE210480 CTCF bigWigs declare exactly hg19 and the whole 12,443-anchor ranking was computed in hg19. Every strength, ratio, rank and percentile in this beat stands. The error is confined to a convenience coordinate quoted for the reader.",
  "AND_THE_COMPARATOR_IS_NOW_ANNOTATED_AT_THE_CORRECT_POSITION": {
    "why_this_is_here": "this beat's own caveat said 'the chr11:8,241,000 comparator has not been annotated: what gene or element it sits in was not looked up'. It has now been looked up - at the corrected position, because the first attempt used the wrong one.",
    "GRCh38 chr11:8,219,453 overlaps": "lncRNA ENSG00000307368, 8,202,758-8,224,376",
    "LMO1 (protein_coding, 8,224,309-8,268,807)": "4,856 bp away",
    "other features within 50 kb": "ENSG00000307401 lncRNA 9,812 bp; ENSG00000307386 lncRNA 12,986 bp; RIC3-DT lncRNA 39,206 bp",
    "THE_COMPARATOR_IS_NOT_INERT": "LMO1 is an established T-lineage oncogene and neuroblastoma susceptibility gene, and the anchor sits 4.9 kb from it inside a lncRNA. So this is NOT the featureless intergenic control the proposal implies. It remains a valid STRENGTH-matched comparator - that was the matching criterion and it holds - but 'a strong anchor that loses more CTCF than KMT2A and has no disease relevance' is not something this annotation supports.",
    "what_is_still_unchecked": "the beat's other caveat stands and is not addressed here: whether this anchor is in a blacklist region or a segmental duplication was not tested, and a strong anchor with a large apparent change is what a blacklist artefact looks like"
  },
  "found_by": "the king, while running this beat's own recommended next step"
}
How this could bite us back
  • A dossier this detailed about one element in retinal pigment epithelium invites the assumption that the element is characterised. It is not: its motif has never been scanned, its methylation has never been read, and its occupancy has never been measured in any blood cell. An intervention argument built on the percentiles here is built on a cell type that cannot give rise to the disease.
  • The 0.396x figure will be quoted as the effect size at this anchor and it should not be, because the four per-replicate ratios span 0.200x to 0.499x. Any dose-response or threshold reasoning about how much CTCF loss is needed for a cytogenetic event is, on these data, reasoning across a 2.5-fold range with no interval.
Still open
  1. Does hg19 chr11:8,241,000 (GRCh38 chr11:8,370,285) amplify or rearrange on KDM3B depletion? It is a strength-matched anchor on the same chromosome that loses MORE CTCF than the KMT2A anchor. If it does not undergo a cytogenetic event, CTCF loss is necessary and not sufficient, and this single locus is the cheapest available test of that - Gray's own copy-number data should cover it.
  2. What is the CTCF motif score and orientation at this element? Four experiments in three cell types now agree on its position to within ~100 bp and not one has produced a motif call. It is a desk job on 550 bp of sequence and it is the first of the material properties that hyp-the-kmt2a-anchor-is-hypersensitive-not-uniquely-targeted needs.
  3. Is this anchor's strength percentile the same in a haematopoietic cell? The strength axis has never been measured outside epithelial and lymphoblastoid cells, and a strong anchor in RPE could be a weak one in a CD34+ progenitor - which would change the whole hypersensitivity argument and is exactly the gap absence-no-human-ctcf-or-cohesin-occupancy-measurement-in-a-blood-cell-at-kmt2a names.
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    Measured from the four unique RPE CTCF bigWigs of GSE210480 at 200 bp genome-wide: the KMT2A exon-11 CTCF anchor's argmax bin is hg19 chr11:118,359,400 = GRCh38 chr11:118,488,685, nine bases from the independently measured HCT116 CTCF summit at GRCh38 chr11:118,488,694; its baseline CTCF strength over a fixed 1,000 bp window is 4.92x the track's own genome mean, the 91.7th percentile of 12,443 control-called anchors whose strength ranges 2.51x to 14.06x; its siKDM3B/control ratio is 0.396x with both siKDM3B replicates below both controls, rank 58 of 1,625 anchors matched within +/-10% of its strength (matched median 0.633) and rank 258 of all 12,443 (median 0.696); and the most CTCF-losing anchor on chr11 is hg19 chr11:8,241,000 = GRCh38 chr11:8,370,285 at strength 4.95x and ratio 0.281x, a closer strength match than any of the eleven hand-picked intervals and a larger response.

    GEO GSE210480 - the four unique RPE CTCF ChIP bigWigs read genome-wide at 200 bp over HTTP range requests with pybigtools 0.3.0, anchors called from the control tracks only, 2026-09-22; pre-registration, code, results and the persisted 12,443-anchor table at analyses/q111-ctcf-anchor-matched-ranking/ · 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
    What this evidence cannot show

    The 200 bp bin means the argmax is a BIN LABEL and not a summit; the nine-base agreement with the HCT116 summit is therefore a coincidence of bin boundaries as much as a measurement, and the honest positional precision is +/-100 bp. The signal-weighted centroid, which is the right estimator at this resolution, is in genome-kmt2a-exon-11-ctcf-summit-rpe-gray and is what should be quoted for position. 'Anchor' here means a local maximum of a deposited ratio track above a fixed fold cut with a both-control-replicates filter - there is no input model, no local background and no FDR, and the Input samples of this series have no deposited file. Strength is a RATIO-track magnitude and is not comparable with raw-coverage magnitudes from other beats. n = 2 versus n = 2, one experiment, no spike-in. hTERT-RPE: epithelial, non-haematopoietic, unrearranged.

    Evidence that argues against this (1)
    • A percentile computed against anchors called by this beat's own rule is a percentile against that rule. A different threshold, window or merging distance would produce a different population and a different percentile, and no external peak set was used as a cross-check - deliberately, so that baseline strength is measured in the same files as the ratio, but the cost is that the denominator of every percentile here is unvalidated against an independently called RPE CTCF peak set. The threshold sweep (2.5x to 5.0x, percentile 3.57-3.58%) bounds one axis of that sensitivity and not the others. Iqbal Lab oncofusion knowledge graph 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    The two siCT control CTCF replicates disagree two-fold at this element - canonical exon 11 reads 9.243x and 4.688x of each track's own chr11 mean, against siKDM3B values of 1.850x and 2.339x - so the four per-replicate siKDM3B/control ratios there span 0.200x to 0.499x; the same spread carries to the summit window (8.553 / 4.349) and the 1,048 bp element (6.120 / 3.213); the whole 90,341 bp KMT2A gene span is flat and essentially unchanged (control 1.002 / 0.987, ratio 0.956); and the pre-declared split-replicate bias control, which calls anchors and measures strength from control replicate 1 and uses control replicate 2 alone as the ratio denominator, places the anchor at the 31.9th percentile of 185 matched anchors with a ratio of 0.586x.

    GEO GSE210480 CTCF bigWigs, per-replicate interval means over six KMT2A intervals normalised to each track's own chr11 mean, with hg19 coordinates from an independently re-run Ensembl lift (constant offset 129,285 bp across six widths from 2 bp to 90,341 bp), 2026-09-22; p5_reproduction_of_the_published_claim and split_replicate blocks of analyses/q111-ctcf-anchor-matched-ranking/results.json · 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
    What this evidence cannot show

    A two-fold spread between two replicates of the same arm is not abnormal for a ChIP fold-enrichment at a strong site and does not make the mean wrong; what it bounds is the PRECISION of the magnitude, and that bound is itself derived from n = 2 and so is imprecise. No confidence interval is quoted because two observations per arm do not support one. The cause cannot be assigned - ChIP efficiency, depth, the fold-over-control denominator or biology - because the Input samples of this series carry no deposited file, so depth, duplicate rate and FRiP are all unrecoverable.

    Evidence that argues against this (1)
    • The replicate spread could be read as undermining the published result, and it does not. Both siKDM3B replicates are below both controls at all three KMT2A intervals, every pairing gives a ratio below 0.5, and the 90 kb gene span is flat - so CTCF falls at this anchor and not across the gene, robustly. Only the claim that the fall is unusually LARGE depends on the pairing, which is precisely the claim the matched ranking was run to test. Iqbal Lab oncofusion knowledge graph 2026
Connected to