Oncofusion

start / graph / ana-q152-the-332-promoter-aff1-only-set-is-below-the-count-a-random-eight-sample-arm-produces

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q152-aff1-only-recurrent-core

The 332-promoter AFF1-only set is BELOW the count a random eight-sample arm produces (null 425)

also called q152 null for the 332

Three hundred and thirty-two gene switches are switched on somewhere in the eight KMT2A::AFF1 leukaemias and nowhere else. Shuffling which eight patients we call the group gives an average of four hundred and twenty-five such switches. The AFF1 group is, if anything, slightly LESS distinctive than a random group of eight.

The same thing, for a scientist

Under the same occupancy rule, 332 of 19,404 promoters are occupied in >=1 of the 8 KMT2A::AFF1 samples and 0 of the 19 non-AFF1 samples. Holding the occupancy matrix fixed and permuting the 8-sample index arm 10,000 times gives mean 425 (SD 89, 5th-95th 288-580, full range 184-763). The observed 332 sits at p_lower = 0.1465 and p_upper = 0.8566. The '332 AFF1-only promoters' figure that anchors the q145-q151 IRX1 line of argument is therefore not an enrichment at all; it is slightly under the chance expectation for any eight of these twenty-seven samples.

Go deeper
Why this matters more than the 17

The 17-promoter core was always going to be a small-number question. The 332 is the figure that has done the work: it is the set IRX1 was found in, and it has been described as 'AFF1-restricted'. Its null had never been computed. It is 425 1.

The direction is the surprise. A partner-exclusive promoter set is produced most abundantly by an arm whose samples have low pass rates or idiosyncratic occupancy; the AFF1 arm contains the highest-pass-rate sample in the deposit (p247, 14.01%) and none of the lowest, so it produces fewer exclusive promoters than a random eight would.

What survives

'IRX1 is occupied only in KMT2A::AFF1 samples' remains a true statement about these data. What does not survive is the implicit 'and that pattern is itself unusual'. The pattern is shared with 331 other promoters, and 425 is what you would expect from any eight of these samples 1. The q151 entry already recorded the first half of this ('the AFF1-restricted pattern is shared with 331 other loci') with its null field left null; this analysis fills it.

Facts and figures
scriptanalyses/q152-the-aff1-only-recurrent-promoter-core/step3.py
statusrun
result summaryObserved 332 AFF1-only promoters; label-permutation null mean 425 (5th-95th 288-580). p_lower 0.1465.
inputs
[
  "C:/Temp/q147part/promoter_matrix.tsv"
]
measurements
[
  {
    "quantity": "count_partner_exclusive_promoters",
    "subject": "promoters occupied in >=1 of the 8 KMT2A::AFF1 samples and 0 of the 19 non-AFF1 samples",
    "value": 332,
    "units": "promoters",
    "n": 19404,
    "universe": "19,404 refGene 2 kb promoters x 27 GSE252378 samples, two-antibody 2.0x plus per-library floor",
    "null": 425,
    "note": "10,000 label permutations; SD 89, 5th-95th 288-580, range 184-763; observed is BELOW the null, p_lower 0.1465"
  }
]
Still open
  1. Is the number of partner-exclusive promoters a monotone function of the index arm's mean pass rate? If so the whole quantity is a depth/enrichment readout wearing a biology label.
  2. What is the equivalent null for each of the other five partner families in this deposit, and does any family beat it?
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    Over the 19,404 x 27 binary occupancy matrix, the observed count of promoters occupied in >=1 of the 8 AFF1 samples and 0 of the other 19 is 332, and 10,000 label permutations of the same matrix give mean 425, SD 89, 5th-95th percentile 288-580, full range 184-763, one-sided p_lower 0.1465 and p_upper 0.8566.

    Q152 label-permutation null for the AFF1-only promoter set · Iqbal Lab oncofusion knowledge graph, beat Q152 · this project 2026
    What this evidence cannot show

    The permutation preserves each sample's pass rate and each promoter's cross-sample correlation but does not preserve partner family structure; a within-family permutation was not run.

    Evidence that argues against this (1)
    • A count below its null is not evidence against partner specificity for any individual promoter. The permutation averages over which eight samples are drawn, and an arm with genuine shared biology could produce FEWER exclusive promoters than a random arm precisely because its members agree with each other - the same mechanism that makes the null high. Iqbal Lab oncofusion knowledge graph et al. 2026
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    The q151 entry records 332 of 6,049 (5.49%) as the AFF1-restricted share and states in its own note that the pattern is shared with 331 other loci, with the measurement's null field left empty.

    finding-q151-the-two-antibody-2x-occupancy-rule-calls-a-tenth-of-all-promoters-occupied (this graph) · Iqbal Lab oncofusion knowledge graph, beat Q152 · this project 2026
    What this evidence cannot show

    Reading another beat's entry, not re-deriving its prose. Its numeric figures were independently re-derived and match to the digit.

    Evidence that argues against this (1)
    • The q151 entry's measurement left its null field empty rather than asserting a wrong null, so describing this as a refutation of q151 overstates it; q151's own note already said the pattern is shared with 331 other loci. The correction is to the reading of that figure elsewhere in the project, not to the entry. Iqbal Lab oncofusion knowledge graph et al. 2026
Connected to