Oncofusion

start / graph / ana-q122-salvaged-the-menin-inhibitors-h3k79me2-effect-is-indistinguishable-in-a-line-with-no-kmt2a-fusion

Analysisweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.

SALVAGED: in K562, which carries no KMT2A fusion at all, a menin inhibitor moves H3K79me2 as much as it does in fusion-bearing murine MLL-AF9 (V 0.8291 vs 0.7734, F 0.1831 vs 0.2500). Both the confirmation and the refutation fail to fire, and the refutation misses by 0.0057 - this is at-the-edge and must not be read as a clean negative

also called the K562 no-fusion control · does the menin inhibitor need a KMT2A fusion to move H3K79me2 · Q122 salvage, arm 2

If a drug aimed at a fusion protein changes a chemical mark on DNA packaging, the obvious question is whether it still does so in cells that have no such fusion - and this compared exactly that, using a leukaemia line with no KMT2A rearrangement. The drug moved the mark in that line by very nearly the same amount as in cells that do carry the fusion, which points away from the change being a consequence of hitting the fusion. The honest caveat is that the pre-set test for 'the effect does not need the fusion' missed its threshold by less than one part in a hundred and a second test also failed to reach its own bar, so this sits genuinely on the fence rather than settling the question either way.

The same thing, for a scientist

**PROVENANCE: coordinator salvage.** Q122 computed this and was killed by the account session limit before writing any node. Figures are read from the beat's committed `results.json`; none was re-derived from raw data by me. **The arm my brief did not ask for, and the most consequential one.** K562 carries **no KMT2A fusion** (CVCL_0004). Under 500 nM VTP-50469 for 7 days against DMSO 7 days, at the mark's own domains: | | V (VTP/DMSO) | G (SGC0946/DMSO) | F | |---|---:|---:|---:| | **K562, no KMT2A fusion** | **0.8291** | 0.0666 | **0.1831** | | murine MLL-AF9, F3, S_FLAG | 0.7734 | 0.0937 | 0.2500 | | murine MLL-AF9, F3, S_K79 | 0.9117 | 0.0675 | 0.0947 | **K562's response sits inside the range of the fusion-bearing murine arms.** Its F of 0.1831 is between the murine 0.0947 and 0.2500. **AND BOTH PRE-REGISTERED CRITERIA FAIL TO FIRE, WHICH IS THE HONEST HEADLINE.** - **C5, "the effect needs a KMT2A fusion"**: fires if V_K562 ≥ 0.9 (i.e. the drug barely touches the mark without a fusion). Observed **0.8291. DOES NOT FIRE.** - **R2, "the drug moves the mark without any KMT2A fusion"**: fires if V_K562 ≤ V_F3(S_FLAG) + 0.05 = **0.8234**. Observed **0.8291. DOES NOT FIRE — by 0.0057.** **A pre-registered refutation missing its threshold by 0.0057 is not a negative result and this entry refuses to report it as one.** The two criteria bracket the observation from opposite sides and neither closes. The defensible statement is narrow: **the menin inhibitor's effect on H3K79me2 in a fusion-free line is close enough to its effect in a fusion-bearing line that a pre-registered test could not separate them**, and the direction of the miss is toward fusion-independence. **Why this matters more than its size.** If the menin inhibitor's H3K79me2 effect does not require a KMT2A fusion, then every statement in this corpus treating that effect as a readout of fusion disruption is describing something else — possibly a general consequence of menin inhibition on chromatin, possibly an indirect proliferation effect over 7 days. Combined with **R4 firing** (the VTP and SGC0946 bin-wise responses do not correlate, rho = −0.026), the mark's movement under a menin inhibitor looks neither fusion-specific nor DOT1L-pathway-mediated.

Go deeper
Why 0.0057 gets its own paragraph

This corpus has a registry row about pre-registered thresholds and a documented case of a verdict flipping across a fixed cutoff on a deviation introduced for an unrelated reason. R2 missing by 0.0057 is that hazard in its purest form, and the temptation to write "the drug moves the mark without any fusion" is exactly what the pre-registration exists to resist.

What would move it by more than 0.0057, all of which are real and none of which were controlled:

  • The arms are not dose- and duration-matched. K562 is 7 days; the murine comparator V_F3 is 48 h. R2's threshold is defined as V_F3 + 0.05, so a comparator measured at a different timepoint sets the bar.
  • Different species, different line, different genome build. K562 is human hg38; the comparator is murine mm10.
  • K562 is a CML-derived erythroleukaemic line with a BCR::ABL1 fusion — "no KMT2A fusion" is not "no fusion", and its chromatin is not a neutral background.

So the right next step is not to re-read this number but to widen the control: a second fusion-free line, and a duration-matched murine comparator. Either could move the verdict across the line in either direction.

Facts and figures
SALVAGE PROVENANCEQ122 computed this and was killed by the account session limit before writing any node file. Figures read from its committed results.json by the coordinator; nothing re-derived from raw data.
lineK562, CVCL_0004, no KMT2A fusion, human hg38
treatment500 nM VTP-50469 7 d vs DMSO 7 d; SGC0946 5 uM for G
V k5620.8291
G k5620.0666
F k5620.1831
why this is reported as at the edgethe two pre-registered criteria bracket the observation from opposite sides and NEITHER closes. A refutation missing its threshold by 0.0057 is not a negative result, and this corpus has a documented case of a verdict flipping across a fixed cutoff on an unrelated deviation.
reads alongsideR4 also fires (rho = -0.026), so the mark's movement under a menin inhibitor looks neither clearly fusion-specific nor DOT1L-pathway-mediated
absence classmeasured-not-analysed
absence class noteA second and third fusion-free line, and a duration-matched murine comparator, are the controls that would settle this. GSE260742 and the wider GEO archive may contain them; nobody has looked. The measurement that exists was made and committed by Q122 and is analysed here for the first time, but the CONTROL that would make it interpretable has not been sought.
comparator murine mllaf9 F3 at S FLAG
{
  "V": 0.7734,
  "F": 0.25
}
C5 the effect needs a kmt2a fusion
{
  "fires_if": "V_K562 >= 0.9",
  "observed": 0.8291,
  "verdict": "DOES NOT FIRE"
}
R2 the drug moves the mark without any kmt2a fusion
{
  "fires_if": "V_K562 <= 0.8234",
  "observed": 0.8291,
  "margin": 0.0057,
  "verdict": "DOES NOT FIRE"
}
unmatched between the compared arms
[
  "duration: K562 7 d vs the murine comparator 48 h - and R2's threshold is DEFINED as V_F3 + 0.05, so the comparator's timepoint sets the bar",
  "species and build: human hg38 vs murine mm10",
  "K562 is CML-derived erythroleukaemic and carries BCR::ABL1 - 'no KMT2A fusion' is not 'no fusion'"
]
Still open
  1. A second and third fusion-free line. With n=1 line the observation cannot be generalised, and K562's erythroleukaemic, BCR::ABL1-positive background is a poor sole negative control.
  2. A duration-matched comparator. R2's threshold is V_F3 + 0.05 where V_F3 is measured at 48 h and V_K562 at 7 d; matching them could move the verdict across the line in either direction, and the miss is only 0.0057.
  3. Does a 7-day exposure in a proliferating line confound the chromatin readout with selection? A shorter K562 timepoint would separate a direct effect on the mark from a shift in the culture's cell-state composition.
Every source behind this entry (1)
  1. E1weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.analysis

    In K562 (CVCL_0004, no KMT2A fusion), 500 nM VTP-50469 for 7 days against DMSO 7 days gives an H3K79me2 within-sample share ratio of V = 0.8291 at the mark's own domains, with SGC0946 giving G = 0.0666 and a resulting fraction of achievable redistribution F = 0.1831 - inside the range of the fusion-bearing murine MLL-AF9 arms (F 0.0947 to 0.2500). The pre-registered confirmation C5 ('the effect needs a KMT2A fusion', fires if V_K562 >= 0.9) DOES NOT FIRE, and the pre-registered refutation R2 ('the drug moves the mark without any KMT2A fusion', fires if V_K562 <= 0.8234) DOES NOT FIRE, missing by 0.0057.

    analyses/q122-h3k79me2-three-arms-with-replicates/results.json, criteria C5 and R2 and arm2_k562_no_kmt2a_fusion; this repository · Oncofusion knowledge graph, Q122 beat (computation) and coordinator (salvage) · internal analysis, reproducible from the repository 2026
    What this evidence cannot show

    CONFIDENCE IS weak AND THE HEADLINE REASON IS THAT NEITHER PRE-REGISTERED CRITERION FIRED - this is an at-the-edge observation, not a result, and R2 missed by 0.0057. Beyond that: the beat died before writing any entry so no agent ever wrote prose against these numbers and I re-derived nothing from raw data. THE ARMS ARE NOT MATCHED - K562 is 7 days and human hg38, the murine comparator that DEFINES R2's threshold is 48 h and mm10, so a comparator at a different timepoint sets the bar. K562 is a CML-derived erythroleukaemic line carrying BCR::ABL1, so 'no KMT2A fusion' is not 'no fusion' and its chromatin is not a neutral background. There is no usable spike-in in GSE260742, so V and G are within-sample SHARE ratios and F is the fraction of achievable REDISTRIBUTION, not of loss. The four K562 files declare an mm10 container while their coordinates are hg38 (the beat's DEVIATION 1).

    Evidence that argues against this (3)
    • The single-line control is the weakness that matters. One fusion-free line cannot distinguish 'the effect does not need a KMT2A fusion' from 'K562 specifically responds this way'. K562 is unusual in ways that bear directly on chromatin readouts - erythroleukaemic, near-triploid, BCR::ABL1-driven - and a second and third fusion-free line are the obvious control. None was run, and with n=1 line the observation cannot be generalised at all.
    • A 7-day exposure in a proliferating line confounds direct chromatin effects with selection and differentiation. Over 7 days VTP-50469 can change the composition of the culture, and a within-sample share ratio computed on a population whose cell-state distribution has shifted measures that shift as much as any direct effect on the mark. The murine arms at 48-72 h are less exposed to this, which means the comparison that R2 rests on is between a short-exposure and a long-exposure measurement.
    • Reading this as evidence against fusion-dependence also has to contend with C5 not firing in the other direction: if the effect were purely fusion-independent, V_K562 should have matched the fusion-bearing arms closely, and at 0.8291 against V_F3 0.7734 the fusion-bearing line actually showed the LARGER effect. The gap is small and in the direction of some fusion-dependence, which is the reading this entry's title deliberately does not take but which the numbers equally permit.
Connected to