Oncofusion

start / graph / ana-q120-every-channel-moves-more-at-the-strongest-bins-and-the-mll1-n-inertness-is-an-aggregate

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q120-h3k79me2-second-line-and-the-mll1-epitope

Stratifying the MOLM13 co-occupied bins by signal amplitude qualifies this beat's own headline: at the strongest tenth the MLL1-N channel falls to 0.5304 and H3K79me2 to 0.7614, so 'MLL1-N does not move' and 'the mark does not move' are both properties of the aggregate - while the ordinal swap that refuted the four-way ordering survives in every stratum

also called the amplitude stratification · the inertness is an aggregate · H3K79me2 does fall at the top bins

Splitting the sites by how much signal they carry shows that at the very strongest ones everything moves, including the two things reported as not moving. The surprising ordering result survives the split; the reassuring flatness does not.

The same thing, for a scientist

Not pre-registered. Run after the criteria were scored because an aggregate ratio over a set spanning two orders of magnitude of signal can hide a gradient, and this one does [E1]. MOLM13, 330 nM VTP-50469, the 467 MLL1-N-and-MENIN co-occupied bins, split by MLL1-N DMSO amplitude: | stratum | n | share of the set's MLL1-N signal | MENIN | DOT1L | H3K79me2 | MLL1-N | input | |---|---:|---:|---:|---:|---:|---:|---:| | strongest 10% | 46 | 0.2046 | **0.0744** | **0.1340** | **0.7614** | **0.5304** | 0.9294 | | strongest 25% | 116 | 0.3964 | 0.1026 | 0.1693 | 0.8099 | 0.6344 | 0.9342 | | middle 50% | 234 | 0.4376 | 0.2187 | 0.2555 | 0.9020 | 0.8978 | 0.9156 | | weakest 25% | 117 | 0.1660 | 0.1620 | 0.2641 | 0.8696 | 0.9500 | 0.9548 | | **all** | **467** | 1.0000 | **0.1657** | **0.2219** | **0.8667** | **0.8020** | 0.9272 | And at the 1,389-bin MLL1-N held set, where this beat and Q118 both report the channel as unmoved: strongest tenth **MLL1-N 0.6919, H3K79me2 0.8453**; weakest quarter MLL1-N 1.0815, H3K79me2 1.0374. **Three things follow, and the first two are corrections to entries in this beat.** 1. **The MLL1-N channel is not uniformly inert.** At the strongest co-occupied bins it falls to 0.53, and at the strongest bins of its own held set to 0.69. The flat aggregate (0.8020 and 0.9438) is produced by the weaker two thirds of each set. `ana-q120-the-mll1-n-channel-is-inert-to-three-decimals-in-both-human-lines` is therefore a statement about the aggregate and carries a pointer to this entry. 2. **H3K79me2 does fall at the strongest sites**, to 0.7614 - which is close to RS4;11's *aggregate* 0.6578. So part of the cross-line H3K79me2 difference may be a difference in how the two sets are distributed over amplitude rather than a difference in the mark's response. **Q118 did not publish a stratified RS4;11 table, so the comparison cannot be completed here**, and that is the cheapest way to strengthen or dismantle this beat's headline. 3. **The ordinal result survives.** In every stratum the order is `MENIN < DOT1L < MLL1-N < H3K79me2`, so the swap that refuted the four-way ordering is not an amplitude artefact. The criterion that fired (F2) fires in all four strata; the criteria about magnitude do not transfer between strata. **The input is flat across strata** (0.9156-0.9548), so none of this is a coverage gradient.

Go deeper
The named loci at the top, annotated only by the series' own RefGene windows

The ten strongest MOLM13 co-occupied bins, per-bin share ratios, gene labels taken only from GSE127507's own deposited ROSE TSS windows so that no external resolver can put a gene on the wrong chromosome 1:

locus (hg19)ROSE geneMLL1-NMENINDOT1LH3K79me2
chr10:65,224,000JMJD1C-AS10.1890.0340.0250.661
chr10:65,280,000REEP30.3220.0550.0240.413
chr15:28,982,000WHAMMP20.2750.0360.0270.537
chr12:93,964,000SOCS20.5580.1170.0430.332
chr1:149,824,000HIST2H2BC0.7560.3450.2971.429
chr4:15,004,000LOC4410090.3910.0380.0660.482
chr13:72,438,000DACH10.2680.0490.0180.639
chr1:51,434,000CDKN2C0.3970.0450.1000.640
chr15:23,208,000WHAMMP30.2260.0460.0110.492
chr9:214,000DOCK80.3640.0300.0440.570

JMJD1C tops the list, as it did for RS4;11 in Q118, and this beat's top-30 table is written to results-arm1.json in full. At these bins the MLL1-N channel loses 24-81% of its share and H3K79me2 loses 33-67% at nine of the ten - so a reader looking only at named loci would conclude the opposite of what the aggregate says, which is why both are reported.

Facts and figures
scriptanalyses/q120-h3k79me2-second-line-and-the-mll1-epitope/stratify-by-amplitude.py
statusrun
result summaryEvery channel moves more at higher amplitude. At the strongest tenth of the co-occupied set MLL1-N is 0.5304 and H3K79me2 0.7614, against aggregates of 0.8020 and 0.8667. The ordinal swap that refuted the four-way ordering holds in all four strata.
cell lineMOLM13
fusionKMT2A::MLLT3
drugVTP-50469 330 nM
qualifies entryana-q120-the-mll1-n-channel-is-inert-to-three-decimals-in-both-human-lines
epigenetic readoutH3K79me2 falls to 0.7614 at the strongest tenth of the MOLM13 co-occupied bins and is 0.8696 at the weakest quarter, against an aggregate of 0.8667
mark writtenH3K79me2
writer of that markgene-dot1l
shown for fusionKMT2A::MLLT3 in MOLM13 only; the equivalent stratification for KMT2A::AFF1 has not been computed by anyone
absence classmeasured-not-analysed
absence class noteThe RS4;11 stratification that would complete this comparison is pure compute on files already public in GSE127507 - GSM3635695/96, 3635693/94, 3635691/92, 3635697/98 - and has never been run. No wet work.
absence class assigned2026-09-23 (Q120)
How this could bite us back
  • An aggregate share ratio over a set spanning two orders of magnitude of signal is a signal-weighted average that can report 'no change' while the sites anybody would name in a figure lose half their signal. Both of this beat's flatness claims are of that kind, and so is Q118's.
Still open
  1. Does RS4;11 show the same amplitude gradient? If its strongest tenth gives H3K79me2 near 0.5 and its weakest quarter near 0.8, then the cross-line difference shrinks to an amplitude-distribution difference and this beat's headline weakens substantially. Four downloads.
  2. Why does everything move more at the strongest bins? A simple regression-to-the-mean account predicts it for a noisy n = 1 measurement, and the flat input argues against a coverage explanation, but no null was computed for the gradient itself.
Every source behind this entry (1)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    Stratifying the 467 MOLM13 MLL1-N-and-MENIN co-occupied 2 kb bins by MLL1-N DMSO amplitude gives, for the strongest 10% (n = 46, carrying 20.46% of the set's MLL1-N DMSO signal), share ratios MENIN 0.0744, DOT1L 0.1340, H3K79me2 0.7614, MLL1-N 0.5304, input 0.9294; for the weakest 25% (n = 117) 0.1620, 0.2641, 0.8696, 0.9500, 0.9548. At the 1,389-bin MLL1-N held set the strongest 10% (n = 138) give MLL1-N 0.6919 and H3K79me2 0.8453. The order MENIN < DOT1L < MLL1-N < H3K79me2 holds in every stratum of both sets.

    What this evidence cannot show

    NOT pre-registered; run after the criteria were scored, so the strata and their boundaries are post-hoc and no threshold is attached to them. The comparison that would matter most - the same stratification of RS4;11 - could not be made, because Q118 stored aggregates only and this beat did not re-bin the RS4;11 files. 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. Bethyl A300-086A, deposited as 'anti-MLL1n', an MLL1 N-TERMINUS epitope. The N-terminus is retained in every KMT2A fusion, so this track is the sum of the fusion and wild-type MLL1 in unknown proportion and CANNOT distinguish them. The immunogen residue range was not verified: three vendor routes were tried this session (fortislife.com two URL forms, bethyl.com/product/A300-086A) and all returned HTTP 404, so 'N-terminus' rests on the depositors' own target label and on the standard fact that MLL-FPs lack the C-terminal SET domain. 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.

    Evidence that argues against this (1)
Connected to