Oncofusion

start / graph / genome-kmt2a-bcr-in-base-positions-three-definitions

Non-codingestablished confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.Q108-h3k9me3-at-the-bcr

The KMT2A breakpoint cluster region in base positions: three incompatible definitions reconciled to one GRCh38 table, with the DCAL frame's three bp identities re-derived and the GRCh37 lift measured per interval

also called the BCR, in base pairs · which 8.3 kb do you mean · chr11:118,482,422-118,489,887

Three entries in this project describe the stretch of KMT2A where the gene breaks, and all three give different exon numbers for what they claim is the same piece of DNA. This converts every version into plain chromosome coordinates so nobody has to guess again.

The same thing, for a scientist

This graph holds **three incompatible exon-based definitions of one breakpoint cluster region** - DCAL/recombinome 'intron 7 to exon 13', Ensembl canonical, and the historic 'exons 5-11' - plus a fourth frame in *mouse* *Kmt2a* and a fifth in Gothe's 'exon 12 BCR'. An exact false claim has already been published in this graph by merging two of them [E2]. Every interval below was derived from **ENST00000534358 in GRCh38** and lifted to GRCh37 by a **per-interval** Ensembl call filtered on `seq_region_name == "11"`, with the offset re-derived rather than assumed: **exactly 129,285 bp, constant across nine intervals from 1,048 bp to the 90,341 bp gene span** [E1]. The three DCAL subregion lengths that fix the frame as canonical + 1 were re-derived and match to the base: canonical e8+i8 = **1,761**, e9+i9 = **679**, e10 start to e12 end = **5,026** [E1]. **The recommended primary definition is DCAL subregions A+B+C = GRCh38 chr11:118,482,422-118,489,887, 7,466 bp**, because it is the only definition whose arithmetic reproduces Meyer's published subregion lengths exactly and because those subregions carry the 3,401-patient distribution. **It is not 8.3 kb** - the commonly quoted figure corresponds to a wider window, and both are tabulated below so the discrepancy is visible rather than silently absorbed.

Go deeper
The table, GRCh38 first
featureGRCh38GRCh37/hg19width
BCR core, DCAL subregions A+B+C (recommended)chr11:118,482,422-118,489,887chr11:118,353,137-118,360,6027,466 bp
BCR wide, DCAL 'intron 7 to exon 13'chr11:118,480,239-118,489,887chr11:118,350,954-118,360,6029,649 bp
CTCF+RAD21+SMC3 elementchr11:118,488,159-118,489,206chr11:118,358,874-118,359,9211,048 bp
Ensembl canonical exon 11chr11:118,488,614-118,488,760chr11:118,359,329-118,359,475147 bp
measured CTCF summit +/-250chr11:118,488,444-118,488,944chr11:118,359,159-118,359,659501 bp
DCAL functional cut point +/-250chr11:118,484,612-118,485,112chr11:118,355,327-118,355,827501 bp
20 kb centromeric flank of the corechr11:118,462,422-118,482,421chr11:118,333,137-118,353,13620,000 bp
20 kb telomeric flank of the corechr11:118,489,888-118,509,887chr11:118,360,603-118,380,60220,000 bp
KMT2A gene, ENST00000534358chr11:118,436,492-118,526,832chr11:118,307,207-118,397,54790,341 bp

Canonical exon boundaries this beat used, GRCh38, so nobody has to refetch them: e6 118,480,174-118,480,238; e7 118,481,715-118,482,092; e8 118,482,422-118,482,495; e9 118,484,183-118,484,314; e10 118,484,862-118,484,975; e11 118,488,614-118,488,760; e12 118,489,792-118,489,887; e13 118,490,129-118,490,249 1.

Two corrections this table forces on entries I do not own.

  1. breakpoint-kmt2a-major-bcr states that its own 'exon 11' is "canonical exon 10, chr11:118,484,862-118,484,993". Canonical exon 10 in ENST00000534358 is chr11:118,484,862-118,484,975, 114 bp. The recorded end is 18 bp too far and the implied width 132 bp instead of 114 bp. The start is right. The error does not affect that entry's conclusions, which turn on the cut point at 118,484,861|862, and it is reported rather than edited 1.
  2. genome-kmt2a-exon-11-ctcf-summit-measured gives the KMT2A GRCh37 span as chr11:118,307,205-118,397,539 with its own caveat that the GRCh38 span was derived by adding the offset. Both spans are defensible and they are not the same object: lifting the GRCh38 span gives GRCh37 chr11:118,307,207-118,397,547, while looking the gene up in the GRCh37 archive gives 118,307,205-118,397,539. The 2 bp and 8 bp differences are gene-annotation differences between two builds, not assembly differences - the assembly offset is constant at 129,285 bp even across this span. That entry's caveat was right to flag the row; the reason is annotation drift, not a bad lift 1.
Why 7,466 and not 8,300, and the rule that follows

The 8.3 kb figure is real and it is a different window. DCAL subregions A+B+C sum to 1,761 + 679 + 5,026 = 7,466, which is exactly canonical exon 8 start to canonical exon 12 end. Extending to DCAL 'intron 7' - canonical intron 6 - gives 9,649 bp. The conventional 8.3 kb sits between the two and corresponds to neither; adding DCAL exon 8 (canonical exon 7) to the core gives 7,844 bp, which is the closest arithmetic candidate. So '8.3 kb' should be treated as an approximate label for the region and never as an interval, and any measurement over 'the BCR' must state which boundary set it used. This beat used the core for every headline number and reported the wide definition alongside.

The rule, stated once for the corpus. Convert to base positions before putting two exon-numbered features in one sentence. This is already a registry rule here; what this entry adds is the table that makes obeying it free, plus the observation that the conversion has to come from a named, versioned transcript - ENST00000534358 - because exon numbering is a property of a transcript and KMT2A has 36 canonical exons and many alternatives.

And a caution on the lift. The 129,285 bp offset was verified constant across nine intervals spanning 90 kb at 11q23.3, and one of the nine was the whole gene. That does not license the same arithmetic elsewhere on chr11, and it does not license lifting a gene-length span in place of looking it up - as the KMT2A gene row above demonstrates.

Facts and figures
elementthe KMT2A major breakpoint cluster region, expressed in base positions in both builds under three incompatible exon-based definitions
transcriptENST00000534358 (KMT2A-213), GRCh38, 36 exons, plus strand
recommended primary definitionDCAL subregions A+B+C = GRCh38 chr11:118,482,422-118,489,887, 7,466 bp
why that oneit is the only BCR definition in this graph whose arithmetic reproduces Meyer's published subregion lengths to the base (1,761 / 679 / 5,026, sum 7,466), and those subregions carry the 3,401-patient breakpoint distribution
grch38 minus grch37 offset bp129285
offset verified on n intervals9
the 8 3 kb figuredoes NOT correspond to either DCAL definition: A+B+C = 7,466 bp, intron 7 to exon 13 = 9,649 bp, and adding DCAL exon 8 to the core gives 7,844 bp. Treat '8.3 kb' as an approximate label for the region and never as an interval.
localizationchromatin, chr11q23.3, within the KMT2A gene body and its 20 kb flanks
subnuclear localizationnot measured and not claimed - this entry is a coordinate reconciliation and asserts nothing about nuclear position
cell typenot applicable - a genome coordinate is cell-type independent. Where this beat MEASURED signal over these intervals the cell type was hTERT-RPE, and that is recorded with the measurements rather than here.
cell statenot applicable to a coordinate; the measurements that used these intervals were in asynchronous proliferating culture 72 h after siRNA transfection
epigenetic readoutnone directly - the intervals defined here are the coordinate frame for the H3K9me1/2/3 and CTCF chromatin measurements in this beat's other entries
shown for fusionNot shown for any fusion. The coordinates describe the intact KMT2A locus in the reference assembly; 98% of KMT2A-rearranged patient breakpoints fall inside the core interval, but no fusion-bearing cell contributed to any coordinate here.
dcal frame identities rederived
canonical e9 + i9 == DCAL…679.0canonical e9 + i9 == DCAL 'exon 10 - intron 10' == 679: 679canonical e8 + i8 == DCAL…1,761canonical e8 + i8 == DCAL 'exon 9 - intron 9' == 1761: 1761canonical e10start..e12en…5,026canonical e10start..e12end == DCAL 'exon 11 .. exon 13' == 5026: 5026
intervals grch38
{
  "bcr_core_dcal_ABC": "chr11:118482422-118489887",
  "bcr_wide_dcal_i7_e13": "chr11:118480239-118489887",
  "element_ctcf_cluster": "chr11:118488159-118489206",
  "exon11_canonical": "chr11:118488614-118488760",
  "ctcf_summit_pm250": "chr11:118488444-118488944",
  "cut_point_pm250": "chr11:118484612-118485112",
  "flank_centromeric_20kb": "chr11:118462422-118482421",
  "flank_telomeric_20kb": "chr11:118489888-118509887",
  "kmt2a_gene": "chr11:118436492-118526832"
}
dcal frame identities rederived — the numbers behind the chart
{
  "canonical e8 + i8 == DCAL 'exon 9 - intron 9' == 1761": 1761,
  "canonical e9 + i9 == DCAL 'exon 10 - intron 10' == 679": 679,
  "canonical e10start..e12end == DCAL 'exon 11 .. exon 13' == 5026": 5026
}
canonical exon boundaries grch38
{
  "6": [
    118480174,
    118480238
  ],
  "7": [
    118481715,
    118482092
  ],
  "8": [
    118482422,
    118482495
  ],
  "9": [
    118484183,
    118484314
  ],
  "10": [
    118484862,
    118484975
  ],
  "11": [
    118488614,
    118488760
  ],
  "12": [
    118489792,
    118489887
  ],
  "13": [
    118490129,
    118490249
  ]
}
corrections reported not edited
[
  "breakpoint-kmt2a-major-bcr records its own 'exon 11' as canonical exon 10 at chr11:118,484,862-118,484,993; canonical exon 10 in ENST00000534358 is chr11:118,484,862-118,484,975 (114 bp, not 132 bp). The end is 18 bp too far. Start correct; conclusions unaffected.",
  "genome-kmt2a-exon-11-ctcf-summit-measured's KMT2A gene-span row differs between a lift (GRCh37 chr11:118,307,207-118,397,547) and a GRCh37 lookup (118,307,205-118,397,539). Both are defensible; the difference is gene-ANNOTATION drift between builds, not a bad lift - the assembly offset is constant at 129,285 bp even across the gene span. That entry's own caveat flagged the row and should now name annotation drift as the reason."
]
How this could bite us back
  • A coordinate table becomes the thing nobody re-checks. Every row here should be regenerated by running lift.py, which asserts the three DCAL identities and the constant offset and fails loudly if Ensembl's annotation moves, rather than copied out of this entry.
  • Recommending one BCR definition as primary will make the other two look wrong. They are not wrong; they are different windows with different provenance, and an entry built on the historic or wide definition should not be silently re-read against the core. That silent re-reading is exactly the class of error that produced the retracted claim this entry cites.
Still open
  1. Does the DCAL frame's canonical + 1 offset hold across all 36 KMT2A exons, or only in this region? Three exact identities fix it between canonical exons 8 and 12. A frame that drifts elsewhere in the gene would not be caught, and the minor BCR and the second USP2-associated BCR are both outside the verified window.
  2. Which transcript does the historic 'exons 5-11' numbering belong to? It is the one frame of the five that this project has never traced to a transcript, and until it is, any entry using it cannot be converted to base positions at all.
  3. Where exactly does Gothe's 'exon 12 BCR' and Canela 2017's mouse Kmt2a numbering land in these coordinates? The mouse frame is explicitly unreconciled in this graph, and a mouse-to-human exon mapping is a desk job nobody has done.
Every source behind this entry (2)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    All nine intervals were derived from Ensembl canonical transcript ENST00000534358 in GRCh38 (36 exons, plus strand, chr11:118,436,492-118,526,832) and lifted to GRCh37 by per-interval Ensembl assembly mapping filtered on seq_region_name == '11', giving a constant offset of exactly 129,285 bp with widths preserved on all nine; the three DCAL subregion identities were re-derived from those exon boundaries as 1,761, 679 and 5,026 bp; canonical exon 10 is chr11:118,484,862-118,484,975 (114 bp); and lifting the GRCh38 gene span gives GRCh37 chr11:118,307,207-118,397,547 against the GRCh37 archive's own annotation of chr11:118,307,205-118,397,539.

    What this evidence cannot show

    The offset is verified over 90 kb at 11q23.3 only and licenses no arithmetic elsewhere. For the 1,048 bp element and the 147 bp exon the mapping endpoint returns TWO mappings with identical coordinates, one on chromosome 11 and one on the alternative scaffold HG122_PATCH; taking the first would silently return a patch coordinate, and the script filters on seq_region_name == '11'. Exon numbering is a property of the transcript: these numbers are ENST00000534358's and a different transcript gives different exon numbers for the same bases. The DCAL frame's canonical + 1 offset is established on three exact identities in this region and is NOT verified across all 36 exons, so a frame that drifts elsewhere in the gene would not be caught by this check.

    Evidence that argues against this (1)
    • A constant offset is necessary but not sufficient for a correct lift - if both assemblies shared an error at this locus the offset would still be constant. The independent biological check is that the lifted exon-11 interval carries 7.0x-chr11-mean CTCF in Gray's RPE tracks while the lifted cut point 3.8 kb away carries 1.3x, a contrast that would not survive a displacement of even a kilobase. NCBI Gene Expression Omnibus 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    Reading 'exon 11' in two numbering frames has already produced a false published claim in this graph: mechanism-h3k9me1-2-ctcf-balance-gates-kmt2a-rearrangement published 'The CTCF site is at the functional cut point of the BCR', which was retracted in cycle 57 when the DCAL frame was shown to be Ensembl canonical + 1 and the two features were found to be 3,752 bp apart.

    What this evidence cannot show

    This is a statement about this graph's own history, recorded in those entries' own amendment and retraction fields, not about the external literature; it is included because the cost of the trap is the argument for this table existing. The Meyer/Marschalek DCAL numbering itself is not in error - the error was in silently reconciling two frames.

    Evidence that argues against this (1)
    • A table of coordinates can itself become the authority nobody re-checks, which is the failure mode one level up from the one it fixes. Every row here should be re-derived by running lift.py rather than copied from this entry, and the script asserts its own invariants so a re-run fails loudly if Ensembl's annotation moves. Ensembl 2026
Connected to