Oncofusion

start / graph / ana-q120-the-tss-table-says-dot1l-stays-in-molm13-and-the-genome-wide-bins-say-it-leaves-with-menin

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

Two instruments on the same cell line give opposite answers about DOT1L under a menin inhibitor: the depositors' own TSS-window table puts DOT1L at 0.6638 beside menin's 0.2513, and 2 kb genome-wide bins put it at 0.2219 beside menin's 0.1657 - and only the genome-wide instrument can see the 86% of the co-occupied set that is not at a TSS

also called the DOT1L instrument disagreement · O1 fires after all

Using the summary table the original authors deposited, the DOT1L protein looks like it stays put when the drug is added. Going back to the raw coverage files and looking at the whole genome instead of just gene starts, it leaves alongside menin. The table can only see a small slice of the places that matter.

The same thing, for a scientist

**The prior on this question was that it would fail.** Q120's pre-registration says so explicitly (prior P2, written knowing the instrument-A answer): *'I expect DOT1L in MOLM13 not to track MENIN, because instrument A already says so (0.6638 against 0.2513) and because MOLM13's partner is AF9, which binds DOT1L directly.'* Criterion **O1** fires anyway. | MOLM13, menin inhibitor, same cells | MENIN | DOT1L | difference | |---|---:|---:|---:| | **instrument A** - the deposited ROSE TSS table, MENIN-DMSO top 1% of 21,461 windows | 0.2513 | 0.6638 | **0.413** | | **instrument B** - 2 kb genome-wide bins, MLL1-N and MENIN co-occupied, n = 467 | 0.1657 | 0.2219 | **0.056** | | instrument B, pure 3-day set (MLL1-N and DOT1L tops), n = 565 | 0.1866 | 0.2669 | 0.080 | | instrument B, at the MENIN held set, n = 1,389 | 0.1602 | 0.1974 | 0.037 | **Criterion O1 fires on the co-occupied set (0.056 <= 0.08) and lands on the band's own edge on the pure 3-day set (0.080).** Either way the two instruments differ by a factor of five to eleven on the same quantity in the same cell line. **Why instrument A is the one that is wrong here, stated as a measurement rather than a preference.** The deposited table is ROSE signal in 6 kb windows centred on RefGene TSSs - 21,461 of them - and is blind to everything else by construction. Of the 467 genome-wide bins where MLL1-N and menin co-occupy in MOLM13, **13.7% fall inside a deposited ROSE window**; 86.3% are invisible to instrument A. Q118 measured the same blindness from the other side and recorded that the deposited table misses 83% of that beat's gained set. **The generalisable part.** A processed summary table deposited by the authors is not a cheaper version of the raw data; it is a different instrument with its own universe, and here its universe excludes seven eighths of the regions the question is about. The disagreement is not subtle - it is the difference between 'DOT1L stays' and 'DOT1L leaves with menin', which are opposite mechanistic claims.

Go deeper
The exposure confound, and why it does not explain the gap

Instrument A's MOLM13 held set is defined by the 4-day MENIN track while DOT1L is a 3-day sample, so one could argue the TSS answer is an exposure artefact rather than a universe artefact. Two things say otherwise.

First, the genome-wide numbers carry the identical exposure mismatch - the same 4-day MENIN pair - and give 0.056, not 0.413.

Second, the pure 3-day set S_d3 (MLL1-N and DOT1L tops, no MENIN track in its definition) gives DOT1L 0.2669 against MENIN 0.1866, a difference of 0.080. Removing the 4-day track from the set definition moves the answer by 0.024, and switching instrument moves it by 0.357.

So the instrument dominates the exposure by fifteen-fold here, and the pre-registered hazard, while real, is not what produced the disagreement.

Facts and figures
scriptanalyses/q120-h3k79me2-second-line-and-the-mll1-epitope/arm1-molm13-ordering.py
statusrun
result summaryDOT1L in MOLM13: 0.6638 by the deposited TSS table, 0.2219 by genome-wide 2 kb bins at the co-occupied set. O1 (menin and DOT1L leave together) fires on the genome-wide instrument against a pre-registered prior that it would fail.
cell lineMOLM13
fusionKMT2A::MLLT3
drugVTP-50469 330 nM
frac of co occupied bins inside a rose tss window0.137
shown for fusionKMT2A::MLLT3 in human MOLM13; the DOT1L antibody is partner-agnostic
absence classmeasured-not-searched
absence class noteThe signal exists in the archive and was unfindable because the SEARCH SPACE excluded it: the depositors' processed table covers 21,461 TSS windows and 86.3% of the regions this question is about are outside them. GSE127507 holds the raw coverage; re-binning is the fix and needs no wet work.
absence class assigned2026-09-23 (Q120)
How this could bite us back
  • If DOT1L leaves with menin genome-wide in MOLM13, the case for adding a DOT1L inhibitor to a menin inhibitor in MLL-AF9 disease is weaker at the occupancy level than the deposited table suggests - but this beat's own H3K79me2 measurement says the MARK is untouched, so the combination rationale survives on the mark and not on the enzyme.
Still open
  1. How many other conclusions in this corpus rest on the GSE127507 ROSE table alone? Q118 used it as instrument A throughout and its blindness is now measured twice, at 83% of one gained set and 86.3% of one co-occupied set.
  2. Is DOT1L's promoter behaviour genuinely different from its behaviour at the fusion's non-promoter sites, or is the TSS universe simply dominated by wild-type COMPASS promoters where menin inhibition does nothing? Splitting the genome-wide bins by ROSE-window membership would answer it directly and was not done here.
Every source behind this entry (1)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    In MOLM13 under 330 nM VTP-50469 the DOT1L within-sample share ratio is 0.6638 at the MENIN-DMSO top 1% of the depositors' 21,461 ROSE TSS windows (menin 0.2513) and 0.2219 at the 467 genome-wide 2 kb bins where MLL1-N and menin co-occupy (menin 0.1657), 0.2669 at the 565-bin pure-3-day set (menin 0.1866) and 0.1974 at the 1,389-bin MENIN held set (menin 0.1602); 13.7% of the 467 co-occupied bins fall inside any deposited ROSE TSS window.

    What this evidence cannot show

    The instrument-A figures are Q118's stored measurement, re-read from its committed results-instrumentA.json; the genome-wide figures are mine. The MENIN pair is a 4-day exposure in both instruments. 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. CST #77087S clone 1W4Z anti-DOT1L cannot distinguish DOT1L held by the fusion's partner moiety from DOT1L in wild-type complexes - and the two lines compared here differ in exactly that moiety, MLLT3/AF9 (direct DOT1L contact) versus AFF1 (indirect, through the AF4-family/SEC axis). Bethyl A300-105A anti-Menin cannot distinguish menin bound to the fusion from menin bound to wild-type KMT2A/KMT2B, nor from menin's non-KMT2A functions.

    Evidence that argues against this (1)
Connected to