Oncofusion

start / graph / exp-q120-h3k79me2-chip-in-a-kmt2a-afdn-line-under-a-menin-inhibitor

Experimentmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q120-h3k79me2-second-line-and-the-mll1-epitope

The one experiment that separates fusion partner from lineage: an H3K79me2 ChIP under a menin inhibitor in ML-2 or any KMT2A::AFDN line, alongside the MLL1n and MENIN samples that already exist for that line and can be re-binned for free

also called the AFDN arm · H3K79me2 in the third partner

Two cell lines gave opposite answers about the histone tag and they differ in two ways at once. A third line, already used in the same published dataset, would separate the two explanations - but the tag was never measured in it.

The same thing, for a scientist

**The confound.** H3K79me2 at the fusion's own sites under VTP-50469 is 0.6578 in RS4;11 (KMT2A::AFF1, B-ALL) and 0.8667 in MOLM13 (KMT2A::MLLT3, AML). Partner and lineage differ together, so two explanations fit equally [E1]. **ML-2 breaks it in one direction.** KMT2A::AFDN (MLL-AF6) in an AML line - AML like MOLM13, and a partner that does not contact DOT1L directly, like AFF1. GSE127507 already deposits ML-2 DOT1L, MENIN, MLL1n pairs and a DMSO input, **and no H3K79me2 sample** [E2]. **The design, minimal version.** ML-2, 330 nM VTP-50469 or DMSO 0.33% for 3 days - matching the existing samples exactly so the new track joins the deposited ones - H3K79me2 ChIP-seq with CST #5427 (the same reagent as the two existing lines, so no reagent switch enters the comparison), plus a VTP input, which ML-2 also lacks. Two replicates. Six libraries. **The pre-registered prediction is in `hyp-q120`'s PARTNER ARM falsifier**: below 0.80 floor-normalised supports the partner explanation, at or above 0.90 refutes it and lineage wins, 0.80-0.90 is the declared indifference band and the magnitude is reported rather than a verdict. **The free part, which should be done first.** The seven existing ML-2 files have never been binned genome-wide by this project - Q118 analysed ML-2 only through the TSS table. Genome-wide DOT1L, MENIN and MLL1n in a third partner tests the three-quarters of the ordering that DID replicate, costs seven downloads, and is a prerequisite for interpreting any new H3K79me2 track at the same coordinates.

Go deeper
Why the free half matters on its own

Q118 flagged ML-2 as carrying the largest rank effect in its beat - a DOT1L anomaly at p = 3.2e-10 - and filed it as a flag, not a finding, because it came from the weakest library in the table and from the TSS instrument.

This beat has now measured that the TSS instrument and the genome-wide instrument disagree by a factor of five to eleven about DOT1L in MOLM13, with 86.3% of the relevant bins outside every TSS window. So the ML-2 DOT1L flag is exactly the kind of claim the instrument comparison says to re-run genome-wide before believing, and doing so is seven downloads with no wet work.

Facts and figures
statusproposed
systemML-2 (KMT2A::AFDN / MLL-AF6), human AML
drugVTP-50469 330 nM, 3 days, DMSO 0.33% control - matched to the existing GSE127507 samples
n libraries estimate6
free prerequisitere-bin the seven existing ML-2 .deduped.tdf files genome-wide at 2 kb - GSM3635674 through GSM3635680 - which gives DOT1L, MENIN and MLL1n in a third partner and tests the part of the ordering that did replicate
readoutH3K79me2 aggregate within-sample share ratio at the MLL1-N-and-MENIN co-occupied bins, floor-normalised, against the hyp-q120 PARTNER ARM thresholds
what it would settlewhether the H3K79me2 divergence between RS4;11 and MOLM13 is a fusion-partner effect or a lineage effect - currently perfectly confounded
epigenetic readoutH3K79me2 in a KMT2A::AFDN line under a menin inhibitor - never measured
mark writtenH3K79me2
writer of that markgene-dot1l
shown for fusionnobody has measured H3K79me2 under a menin inhibitor in KMT2A::AFDN; it is measured in KMT2A::AFF1 and KMT2A::MLLT3
absence classunclassified-not-checked
absence class noteThe sample is certainly absent from GSE127507, but no GEO-wide or literature query has been run for an H3K79me2 ChIP in any KMT2A::AFDN line under any menin inhibitor, so whether this needs wet work or only a better search is unknown. Honest default; the query is the first step and costs minutes.
absence class assigned2026-09-23 (Q120)
assays
[
  "H3K79me2 ChIP-seq (CST #5427, the same reagent as the two existing lines)",
  "a VTP-50469 input, which ML-2 currently lacks"
]
How this could bite us back
  • If the AFDN line behaves like MOLM13 rather than like RS4;11, the partner hypothesis dies and the surviving explanation - lineage - is the more damaging one, because it says a chromatin mechanism measured in one cell type does not transfer to another cell type of the same disease.
Still open
  1. Would a second AF4-family line be a better control than ML-2? RS4;11 is the only KMT2A::AFF1 line in this series and a second one - SEM or MV4;11 - would test replication within a partner before testing across partners, which is arguably the prior question.
  2. Is ML-2's fusion really KMT2A::AFDN? The identification comes from the series overall design and this beat did not verify it against a karyotype or a cell-line database record.
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    H3K79me2's aggregate within-sample share ratio at the MLL1-N-and-MENIN co-occupied 2 kb bins under 330 nM VTP-50469 is 0.6578 in RS4;11 and 0.8667 in MOLM13, while MENIN, DOT1L and MLL1-N agree between the two lines within 0.035, 0.022 and 0.029.

    What this evidence cannot show

    This is an experiment proposal; the ML-2 H3K79me2 measurement has not been made by anyone. CST #5427 anti-H3K79me2 reads a histone MARK, i.e. a consequence. It cannot distinguish halted deposition from active demethylation from replicative dilution, and it cannot attribute any of them to the fusion. NO SPIKE-IN NORMALISATION. All 26 GSE127507 records state a Drosophila spike-in was run, measured, and deliberately not used, and no scale factor is deposited; no GSE260742 record contains spike-in vocabulary at all. Every cross-condition number here is therefore a WITHIN-SAMPLE share or rank, which cancels the unknown per-sample scale factor exactly. No raw fold change is reported as a biological quantity anywhere in this beat. n = 1 per antibody per condition per line in both series. No number in this beat carries a spread and none is precise.

    Evidence that argues against this (1)
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    GSE127507 deposits for ML-2 a DOT1L pair (GSM3635674/75), a MENIN pair (GSM3635676/77), an MLL1n pair (GSM3635678/79) and a DMSO input (GSM3635680), all at 3 days and 330 nM, and no H3K79me2 sample; ML-2's fusion is named MLL-AF6 (KMT2A::AFDN) in the series overall design.

    What this evidence cannot show

    Read directly from the records.

    Evidence that argues against this (1)
Connected to