Oncofusion

start / graph / ana-q141-the-ha-stratification-separates-the-strata-sixty-two-fold-and-the-authors-gene-list-is-the-top-one-percent

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q141-menin-at-fusion-targets

The stratification passes its own positive control: the fusion tag separates the two strata 62-fold

Before believing any result that splits genes into 'where the fusion was' and 'where it was not', you have to check that the split actually finds the fusion. It does: the fusion's own tag drops to 1% of its starting level in the first group and stays at 61% in the second.

The same thing, for a scientist

Pre-registered gate. Median HA (MLL::AF9) raw dVHL/DMSO TSS ratio 3 h after 500 nM dTAG-VHL is **0.0099** over the authors' 200 target genes (n=200) and **0.6106** over the bottom half of the HA-occupancy ranking (n=8,662) - a 61.7-fold separation. The second 3 h batch (190814) gives 0.028 at target genes, and the 24 h arm 0.055. Independently, the authors' 202-gene target list maps to 200 genes in my HA ranking at **median rank 114.5 of 17,343 - the 0.7th percentile** - so their gene list and an unsupervised cut on HA signal identify the same genes. Every stratified figure in this beat is conditional on this control, and it passes.

Go deeper
Why this had to be computed first

A stratified analysis whose stratification variable does not identify the thing it names produces a clean, confident, meaningless table. The pre-registration made this the gating computation and declared the whole analysis uninformative if HA's own stratum contrast were small 1.

The residual 0.61 in the non-occupied stratum is not fusion

At genes the fusion never occupied, the HA ChIP measures background, and a background-over-background ratio near 0.6 reflects library depth and the absence of a spike-in, not fusion retention. The informative quantity is the 62-fold separation, not either level.

Scale and limits

No spike-in exists anywhere in GSE173599, so every number is a within-sample share, and the treated arms are scaled by a 21-gene housekeeping panel whose own raw 3 h H3K27ac ratio averages 0.809 and is forced to 1. A housekeeping normalisation DEFINES the global change to be zero, so only the DIFFERENCE between strata is measurable; a uniform genome-wide loss of any of these proteins would be invisible here. n = 1 per arm for every channel except HA, which has two independent 3 h batches. The cells are HCB1, transformed by KMT2A::MLLT3 (MLL::AF9) - NOT the index fusion of this project, KMT2A::AFF1 - and a prior beat in this corpus found that per-gene ordering transfers between MLLT3 and AFF1 lines while magnitude does not.

Facts and figures
scriptanalyses/menin-retention-by-ha-occupancy/analyze_tss.py
statusrun
result summaryPositive control PASSES: HA ratio 0.0099 (occupied) vs 0.6106 (non-occupied), 61.7-fold. Authors' target list at HA-rank percentile 0.7.
peak width ratio noteThe companion peak-based HA positive control has treated/control median peak width ratios of 2.79 (batch 1909x2), 1.93 (190814) and 2.18 (1912x2) - all above the pre-registered 1.3 trigger, which is exactly why the signal instrument and not the peak instrument gates this beat.
inputs
[
  "GSE173599_TSS_signal_HA_ChIPs.xlsx",
  "GSE173599_ChIPseq_TSS_Figure3_revisions.xlsx"
]
How this could bite us back
  • If a trial used an HA- or fusion-directed occupancy assay as a pharmacodynamic readout, a 62-fold stratum difference means the answer depends entirely on which genes are sampled; a genome-wide summary would read as a 40% effect when the on-target effect is 99%.
Still open
  1. What is the HA background ratio at genes with no fusion when the library is depth-matched? Without that, the 0.61 non-occupied figure cannot be interpreted as anything but a contrast denominator.
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    Median HA dVHL/DMSO TSS ratio is 0.0099 at the authors' target genes and 0.6106 at the bottom half of the HA ranking, a 61.7-fold separation.

    What this evidence cannot show

    HA columns in this workbook are RAW (not housekeeping-normalised), so they are not on the same scale as the MENIN/DOT1L channels and are used here only as a within-channel contrast. n=1 per arm per batch, but two independent 3 h batches exist and agree in direction (0.0099 and 0.0284).

    Evidence that argues against this (1)
    • A near-total collapse measured without a spike-in cannot distinguish 99% degradation from 99% loss of immunoprecipitable epitope; the dTAG system degrades the protein, but the ChIP reads the HA tag. Olsen SN et al. 2022
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    The authors' published MLL::AF9 target-gene list sits at the 0.7th percentile of an unsupervised HA-occupancy ranking, so the two definitions of 'fusion-occupied' agree.

    What this evidence cannot show

    Both the list and the HA ranking come from the same cells and the same laboratory, so this is a consistency check between two derivations, not independent replication.

    Evidence that argues against this (1)
    • Both the gene list and the HA ranking derive from the same experiments in the same laboratory, so their agreement could reflect shared methodology rather than a robust definition of a fusion target gene. Olsen SN et al. 2022
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    The dTAG-VHL system degrades the FKBP12(F36V)-tagged MLL::AF9 fusion and the paper's central claim rests on that degradation being rapid and near-complete.

    MLL::AF9 degradation induces rapid changes in transcriptional elongation and subsequent loss of an active chromatin landscape · Olsen SN, Godfrey L, Healy JP, Choi YA, Kai Y, Hatton C, Perner F, Haarer EL, Nabet B, Yuan GC, Armstrong SA · Molecular Cell 82(6):1140-1155.e11 2022
    PMID 35245435DOI
    What this evidence cannot show

    Degradation efficiency in the paper is shown by western blot and by the HA ChIP itself; both are within-sample measures.

    Evidence that argues against this (1)
    • A degron tag on the fusion may itself perturb fusion function before any degrader is added, so the DMSO arm is not equivalent to an untagged parental line. Olsen SN et al. 2022
Connected to