Oncofusion

start / graph / exp-spike-in-normalised-histone-chip-under-a-menin-inhibitor

Experimentmoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q-same-condition-noise-floor

The missing reagent is a spike-in, not a better statistic: a spike-in-normalised H3K4me3, H3K79me2 and H3K27me3 ChIP-seq at 72 h of a clinical menin inhibitor in a KMT2A-rearranged human line, which is the only design that can tell a genuine genome-wide mark redistribution from the compositional shadow of a peak-localised change

also called the spike-in experiment · absolute versus relative mark change · what within-sample share cannot see

Every chromatin result in this thread is a percentage of that sample's own total, which means a change that happens everywhere at once is completely invisible. Adding a fixed amount of foreign chromatin to each sample before the measurement gives an absolute yardstick and separates 'the marks moved from here to there' from 'there is simply less of the mark'.

The same thing, for a scientist

**Why.** Cycles 101-104 and this beat all compute WITHIN-SAMPLE shares, because GSE260742 has no usable spike-in - established across 419,824 characters of the deposit with zero hits - and GSE299100's submitter-processed tracks document no normalisation. **A within-sample share cannot see a uniform global change.** Consequently the finding that three histone marks rise at D1 and fall monotonically to D10 under VTP-50469 has two readings that no data in this project can separate: a genuine redistribution, or the arithmetic shadow of a peak-localised gain with shares summing to 1. **Design.** KMT2A-rearranged human line (MV4;11 for KMT2A::AFF1, ideally SEM as well since MV4;11 presents as acute MONOCYTIC with FLT3-ITD rather than the infant B-ALL the project studies). Vehicle and a clinical menin inhibitor (revumenib or ziftomenib) at a dose giving the published target-engagement effect, 72 h to match the deposited timecourse. **Three biological replicates per arm** - the whole reason this question is open is that n = 1 leaves no scale but a borrowed floor. ChIP for H3K4me3, H3K79me2 and H3K27me3 plus matched inputs, with a fixed quantity of *Drosophila* chromatin added per cell equivalent before immunoprecipitation (ChIP-Rx; a spike-specific antibody such as anti-H2Av may be used in addition to measure spike recovery, which is not part of the original ChIP-Rx protocol). **Readouts.** (i) Absolute per-decile mark signal on the parental fusion universe, spike-normalised. (ii) Total genome-wide mark signal per cell equivalent. (iii) The same within-sample share profile, so the new data can be placed beside the existing tables. (iv) A same-condition vehicle-versus-vehicle floor for each mark - the quantity this project has never had for a histone mark. **Controls.** Vehicle-versus-vehicle pairs within the replicate set supply the floor directly. A DOT1L inhibitor arm would separate 'any transcriptional perturbation' from 'menin inhibition specifically'; it is desirable and not required.

Go deeper
What each outcome would mean
absolute D10 mark signaltotal mark signalreading
falls >10%changes <10%real redistribution - the genome-wide hypothesis is supported
flatfalls >25%compositional shadow - the share profile was an artefact of a global loss, and the peak-stratum results in this thread need re-reading too
fallsfalls proportionallyambiguous; the drug reduces the mark everywhere including the bulk, which is a third and simpler mechanism nobody has proposed

The second row is the one that would do the most damage, and it is the reason this is worth doing rather than assuming.

Why a public re-analysis cannot substitute

Two cheaper steps come first and are already listed as falsifiers on the hypothesis - the three-mark per-bin correlation across deciles, and re-cutting the strata on an H3K4me3 reference. Both are computable from bins on disk and neither needs a download. They can refute the redistribution reading. Neither can establish it, because both are themselves within-sample-share analyses and inherit the blindness. Only an absolute measurement closes it.

Facts and figures
designHuman KMT2A-rearranged line (MV4;11 for KMT2A::AFF1; SEM as a B-lineage comparator), vehicle versus a clinical menin inhibitor (revumenib or ziftomenib) at a target-engaging dose, 72 h, three biological replicates per arm. ChIP for H3K4me3, H3K79me2 and H3K27me3 plus matched inputs, with Drosophila chromatin spiked per cell equivalent before immunoprecipitation (ChIP-Rx), reporting every quality-control step named in Patel et al. 2024.
readoutSpike-normalised absolute per-decile mark signal on the parental fusion universe; total genome-wide mark signal per cell equivalent; the matching within-sample share profile for comparability with cycles 101-104; and a same-condition vehicle-versus-vehicle floor per mark, which this project has never had.
controlsVehicle-versus-vehicle pairs inside the replicate set give the floor directly. Input arms per condition. A DOT1L-inhibitor arm to separate menin inhibition from any transcriptional perturbation is desirable, not required.
cost tiermed
timeline8-12 weeks for a lab with ChIP-seq in hand
decides between
[
  "hyp-the-menin-inhibitor-chromatin-response-is-a-genome-wide-redistribution-and-not-a-peak-phenomenon",
  "the compositional-shadow alternative stated inside that hypothesis"
]
cheaper steps first
[
  "Three-mark per-bin correlation across deciles D2-D10, zero downloads, bins already on disk",
  "Re-cut the strata on the H3K4me3 DMSO track instead of FLAG DMSO, zero downloads",
  "Screen the 112 menin-inhibitor GEO series from cycle 100 for an existing spike-in histone deposit",
  "Consider siQ-ChIP (PMID 36945621) as a spike-in-free alternative; it also tests whether the ~9% peak floor is an antibody-concentration effect"
]
How this could bite us back
  • If the experiment shows a large global mark loss, then the menin inhibitor is depleting chromatin marks genome-wide, which is a direct toxicity argument in normal haematopoietic cells and would reframe differentiation syndrome and the class's on-target effects.
  • If it shows a true redistribution, the drug moves marks into the high-signal bulk - regions that include the housekeeping and lineage programmes of a normal cell - which is a different but equally real off-tumour surface.
  • A null result (no absolute change anywhere) would mean the entire decile profile in cycles 103-104 is compositional bookkeeping, and several entries in this corpus would need their mechanism clauses withdrawn rather than qualified.
Still open
  1. Would a spike-in experiment in MV4;11 transfer to infant B-ALL? MV4;11 is KMT2A::AFF1 but monocytic and FLT3-ITD; a SEM arm costs little and covers the lineage objection, which is why it is in the design.
  2. How large a global change would have to exist for the compositional reading to explain the observed decile profile? That is calculable from the existing tables and has not been calculated - doing so first would say whether the required global change is biologically plausible before any wet work is commissioned.
Every source behind this entry (2)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    ChIP-Rx - ChIP with a defined quantity of a reference exogenous (Drosophila) genome - is an established method for genome-wide QUANTITATIVE comparison of histone modification status across cell populations, introduced because 'the lack of an empirical methodology to enable normalization among experiments has limited the precision and usefulness of this technique', and it 'enables the discovery and quantification of dynamic epigenomic profiles across mammalian cells that would otherwise remain hidden using traditional normalization methods', demonstrated specifically for measuring epigenomic changes following chemical perturbations.

    Quantitative ChIP-Seq Normalization Reveals Global Modulation of the Epigenome · Orlando DA, Chen MW, Brown VE, Solanki S, Choi YJ, Olson ER, Fritz CC, Bradner JE, Guenther MG · Cell Reports 2014
    PMID 25437568DOI
    What this evidence cannot show

    Abstract and MeSH read, not the full text. The founding demonstration is a chemical perturbation of lysine methylation in a human leukaemia line (MeSH: Jurkat Cells, Benzimidazoles, Histones, Lysine, Methylation, Drosophila melanogaster) - NOT menin inhibition in KMT2A-rearranged leukaemia, so the magnitude of the problem in this system is inferred rather than measured. The exact compound was not read off the paper and is not asserted here. ChIP-Rx also assumes constant cell number per sample and constant spike recovery through the IP, both of which drift.

    Evidence that argues against this (2)
    • Spike-in normalisation is itself unreliable when implemented loosely. A 2024 Nature Biotechnology analysis - titled, in the authors' own words, 'The Wild West of spike-in normalization' - shows that published implementations routinely omit critical quality-control steps and deviate from the documented procedure, that the approach reduces genome-wide data to a single scalar and is therefore 'particularly vulnerable to errors in implementation', and that misuse 'can create erroneous biological interpretations'. So a single spike-in experiment cannot be treated as definitive, and the quality-control steps in that paper's guidelines must be reported alongside the result. Patel LA et al. 2024
    • There is a spike-in-FREE route to an absolute scale: siQ-ChIP derives a quantitative scale from the physical parameters of the IP itself, and its authors additionally show that a histone mark's apparent genomic distribution depends on antibody concentration - which is directly relevant here, because the same-condition floor measured in this beat is ~9% at the peaks and antibody titration is one candidate cause of it. If siQ-ChIP is used instead, the experiment needs no Drosophila chromatin but does need the IP parameters recorded. Kupai A et al. 2023
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    No spike-in exists in GSE260742, the deposit every chromatin-mark result in this thread rests on, and all ratios there are therefore within-sample relative quantities in which a uniform global change is invisible.

    What this evidence cannot show

    An internal prior finding, re-read rather than re-run by this beat. GSE299100 has not been searched for a spike-in at all; its supplementary tracks are submitter-processed .tdf with no documented normalisation, so the assumption is carried over rather than established.

    Evidence that argues against this (1)
    • If any public menin-inhibitor histone ChIP deposit already carries a spike-in, this experiment is unnecessary and the question is answerable this week. A GDS search over menin-inhibitor terms returned 112 series in cycle 100 and only GSE299100 was examined; none was screened for spike-in. source 2026
Connected to