Oncofusion

start / graph / ana-q191-the-index-fusion-discards-the-siah-degron-and-three-entries-said-it-keeps-it

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.

The index fusion DISCARDS the SIAH degron: the junction is at AFF1 residue 347, the degron is at human 282-304, and three entries had it the other way because they measured a murine motif against an exon-numbering junction

also called AFF1 degron lost not retained · junction 347 resolves the AFF1 coordinate dispute · subtraction hypothesis reinstated for AFF1

The cancer protein is built by joining half of one gene to part of another. A short tag that normally tells the cell to destroy the second protein turns out to sit in the piece that gets thrown away, so the cancer protein has no such tag. Three entries in this project said the opposite.

The same thing, for a scientist

Q187 asserted that KMT2A::AFF1 retains the AFF-family SIAH degron, making it 'a handle rather than an escape'. It does not. Two coordinate errors compounded: the junction was taken as AFF1 codon ~53-54 from Ensembl exon-4 numbering, and the degron was taken as residues 268-290 without noting those are MURINE Af4 coordinates. Corrected, both move the same way. The junction sits at AFF1 residue 347 - QTYSNEVH occurs exactly once in P51825, and the corpus's own siMA6 reconstruction shows both chimeras resuming at VDFK|QTYSNEVHCVEEILK, so the AFF1 acceptor is constant across breakpoints. The degron maps by 22/23 homology to human 282-304, motif PTAYVRP unique at 290-296, robotic-equivalent V294. The element therefore ends 43 residues N-TERMINAL to the junction and is absent from the oncoprotein. This reinstates the subtraction hypothesis that `subtraction-aff1-control` recorded as refuted at 'strong' confidence, and it resolves the junction-codon question that `protein-aff1-domains` called the largest unresolved coordinate in its beat.

Go deeper
Two wrong coordinates that happened to cancel into a confident answer

subtraction-aff1-control reasoned: the degron is at 268-290, the junction is at ~53, 268 - 53 = ~215 residues of margin, therefore retained, therefore the subtraction hypothesis is refuted. It graded the result strong and stress-tested it: 'far larger than any plausible alignment error'.

The stress test guarded the wrong term. The alignment was never in doubt; the junction was. And a second error ran alongside it - 268-290 are murine Af4 coordinates, quoted into a human junction calculation without conversion. The entry half-noticed this ('Human AFF1 differs slightly in length, so 268-290 should be treated as approximate') and then dismissed it using the very margin that the bad junction had manufactured.

This is the failure mode where a caveat names the gap and the prose walks past it.

The junction: 347, by two independent routes, across both breakpoints
  • Sequence. QTYSNEVH occurs exactly once in UniProt P51825 (1,210 aa), at residue 347. Not twice, not approximately - once.
  • mRNA. ana-sima6-is-junction-spanning-and-it-pins-sems-breakpoint independently reconstructed the junction peptide VDFK|QTYSNEVHCVEEILK from the siMA6 siRNA and recorded that 'the AFF1 acceptor is constant and the KMT2A donor is not. Both chimeras resume at QTYSNEVHCVEEILK'. So 347 holds for the intron-11 and intron-9/10 breakpoints alike.
  • Exon structure. protein-aff1-domains had already derived, from ENST00000307808, that 'exon 4' in that transcript is codons 347-361 (17 + 121 + 900 = 1,038 nt = exactly 346 complete codons). Exon arithmetic and protein sequence agree to the residue.

The three competing readings collapse to one physical exon boundary, described in three coordinate systems. Only the codon-54 reading is refuted - and it is refuted by sequence, not by preference.

The degron in human numbering, and the 43-residue verdict
sequencemotifrobotic residue
mouse Af4 268-290KSAAMQQKPTAYVRPMDGQDQAPPTAYVRP 276-282V280
human AFF1 282-304KSVAMQQKPTAYVRPMDGQDQAPPTAYVRP 290-296V294

22 of 23 residues identical, offset +14, and PTAYVRP is unique in human AFF1. The element ends at 304; the junction is at 347. The whole degron lies 43 residues inside the discarded AFF1 leader.

fusion-kmt2a-aff1 already said so without naming the degron - it lists AFF1 1-346 as lost, including the cyclin T1-binding sequence at 1-308. It was corrected to that position on 2026-09-20 and was right; Q187 overrode a sequence-verified field with an exon-numbering inference.

What changes, and what does not

Reinstated. The subtraction hypothesis for AFF1: KMT2A::AFF1 may be stabilised by virtue of a deletion. subtraction-aff1-control is now superseded, and the synthesis entry's 'AFF1 control came through the audit intact' no longer holds - the one plausible hidden subtraction it looked for is real.

A clean dichotomy appears. The reciprocal AF4·MLL retains AFF1 1-346 and so keeps the degron, yet Bursen and colleagues report it escapes proteasomal degradation anyway. Both products evade SIAH turnover; only the forward one does it by losing the motif. That makes the reciprocal the better test of whether the degron is functional at all.

Not established. That SIAH turnover matters to either product. Nobody has blotted either under SIAH manipulation. 'The fusion discards a degron' is a coordinate fact; it is not a demonstrated mechanism, and the temptation to read it as one is what produced the error in the first place - in the opposite direction.

Corrected by the Q194 sweep: a fourth entry already had it right

This entry was published saying three entries had the degron backwards and that fusion-kmt2a-aff1 was the one that was right. The sweep for murine coordinates found a fourth entry that was also right, and more completely so.

mechanism-the-ptefb-binding-module-of-aff1-travels-on-the-reciprocal-product uses the 347 junction throughout and lists, under reciprocal_product_carries, "AFF1 1-346, NHD + ALF, cyclin T1-binding sequence 1-308, Siah-binding degron 268-290" — that is, it had already placed the degron on the reciprocal product and off the forward fusion. It even hedged the number it was quoting: "268-290 (as recorded elsewhere in this graph)".

So the corpus held the correct answer in three places, not one, and the error was localised to the subtraction-* cluster and the entry that inherited it. The one imprecision in that entry was species labelling, not biology — it quoted the murine 268-290 for a human product — and the conclusion is unaffected because the reciprocal carries AFF1 1-346, which contains the element on either numbering. That has now been labelled.

The method lesson, which is the durable part: Q191 found the error by following one claim and reported that one entry had been right. A mechanical sweep over the coordinate itself found a second correct entry that the reasoning had walked straight past. Following a claim finds the entries that cite each other; sweeping the value finds the ones that do not.

Facts and figures
junction residue P51825347
junction peptideVDFK|QTYSNEVHCVEEILK
margin residues degron end to junction43
degron in forward fusionfalse
degron in reciprocal producttrue
prior claim withdrawndegron retained; subtraction hypothesis refuted on coordinates (confidence: strong)
degron murine
{
  "ref": "O88573",
  "element": "268-290",
  "motif": "PTAYVRP 276-282",
  "robotic": "V280A"
}
degron human
{
  "ref": "P51825",
  "element": "282-304",
  "motif": "PTAYVRP 290-296",
  "robotic_equivalent": "V294",
  "alignment_identity": "22/23",
  "offset": "+14"
}
entries corrected
[
  "mechanism-q187-the-aff-degron-is-the-family-dosage-switch-and-three-paralogue-phenotypes-are-stabilisations",
  "subtraction-aff1-control",
  "subtraction-synthesis-how-general",
  "protein-aff1-domains"
]
entry that was right all along
[
  "fusion-kmt2a-aff1 - junction 347, AFF1 1-346 listed as lost, verified by peptide match",
  "mechanism-the-ptefb-binding-module-of-aff1-travels-on-the-reciprocal-product - independently used junction 347 AND assigned the Siah degron to the RECIPROCAL product, i.e. reached this entry’s conclusion before it; found by the Q194 sweep, not by Q191",
  "protein-aff1-domains - derived codons 347-361 for exon 4 under ENST00000307808, the same residue by a different route, though it filed the question as unresolved"
]
Still open
  1. Is KMT2A::AFF1 stabilised because it lacks the degron? The subtraction hypothesis is reinstated and untested; the comparison is fusion versus AFF1-leader-restored construct.
  2. Does SIAH bind human AFF1 at 282-304? Assigned by homology here. The corpus should stop quoting murine coordinates as human ones - this cycle's error came from exactly that.
  3. How many other entries carry murine Af4 coordinates as though they were human? This cycle found three by following one claim; nobody has swept for the pattern.
  4. Does the 347 junction hold across the full KMT2A::AFF1 breakpoint distribution, or only the two breakpoints this graph models? subtraction-aff1-control raised this and it is still open.
Every source behind this entry (4)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    The peptide QTYSNEVH occurs exactly once in the UniProt P51825 canonical AFF1 sequence (1,210 aa), beginning at residue 347; residues 340-360 read MPSQSVEQTYSNEVHCVEEIL.

    What this evidence cannot show

    Fetched and string-matched 2026-09-24. Uniqueness is a property of the canonical sequence; a non-canonical isoform was not tested. This establishes where QTYSNEVH is, not that QTYSNEVH is the junction - that comes from E2.

    Evidence that argues against this (1)
    • P51825 is 1,210 aa while the Ensembl canonical AFF1 protein is 1,218 aa, an 8-residue offset this corpus has recorded before. Residue 347 is therefore reference-specific, and the Ensembl-numbered equivalent is near 354-355 - which is itself the start of the 45-nt exon under ENST00000395146 numbering. The physical boundary is the same; the integer is not portable between references. UniProtKB 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.internal

    The KMT2A::AFF1 junction peptide is VDFK|QTYSNEVHCVEEILK and the AFF1 acceptor is constant across both modelled breakpoints, only the KMT2A donor differing - established in this corpus by reconstructing the siMA6 siRNA target across the join.

    What this evidence cannot show

    CORPUS-INTERNAL. Not re-derived this cycle from a patient junction read. Its weight comes from being independent of the protein-sequence route in E1 and from roughly twenty entries using QTYSNEVH as the junction peptide, not from having been re-verified here.

    Evidence that argues against this (1)
    • Rarer KMT2A::AFF1 breakpoints with a different AFF1 acceptor exist in principle; the constancy claim is across the two breakpoints this graph models, not across the whole breakpoint distribution. `subtraction-aff1-control` itself flagged that the margin should be recomputed over the breakpoint distribution and it never was.
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    The Siah-binding element of Af4 is residues 268-290 with a PxAxVxP consensus and the robotic-mouse V280A lies inside it; in UniProt O88573 (AFF1_MOUSE) residues 268-290 read KSAAMQQKPTAYVRPMDGQDQAP, the motif PTAYVRP occupies 276-282, and residue 280 is indeed V - so the published coordinates are murine and internally consistent.

    What this evidence cannot show

    The 268-290 assignment and V280A phenotype are from Oliver et al. (PMID 15221006); what was verified here is only that those coordinates land on the stated motif in the murine reference. The full text was not re-opened this cycle.

    Evidence that argues against this (1)
    • Oliver and colleagues mapped Siah binding on Af4 peptides in a non-haematopoietic context (cerebellum, Purkinje cells). Whether the same element governs turnover of a KMT2A fusion in a B-lymphoid progenitor is not established by that work, so the element's relevance to the index fusion rests on conservation rather than measurement. PLoS Biology 2004
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    The murine element aligns 22/23 to human AFF1 P51825 residues 282-304 (KSVAMQQKPTAYVRPMDGQDQAP), the PxAxVxP motif PTAYVRP is unique in human AFF1 at 290-296, and the robotic-equivalent valine is V294; the element therefore ends 43 residues N-terminal to the junction at 347 and is absent from KMT2A::AFF1.

    What this evidence cannot show

    The human degron position is assigned BY HOMOLOGY, not measured - no SIAH-binding map exists for human AFF1. The alignment is strong (22/23, motif exactly conserved, single occurrence) but it is still an inference, and this corpus has been wrong before by treating a confident inference as a coordinate.

    Evidence that argues against this (1)
    • Even granting the coordinates, losing a degron does not entail stabilisation. This corpus records that the reciprocal AF4-MLL product escapes proteasomal degradation while CONTAINING the AF4 N-terminus - so in the one AFF1-containing product where retention and turnover have both been assessed, retention did not predict degradation. Loss need not predict stabilisation either.
Connected to