Oncofusion

start / graph / absence-q135-h3k27me3s-declared-second-channel-in-gse260742-does-not-share-a-genome-with-its-first

Analysismoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q135-the-clean-pair-and-the-two-uncomputed-mark-channels

GSE260742 declares an H3K27me3 second channel in its MLLAF9 Control arm, and no floor can be computed from it: all three rep1/rep2 pairs agree about only 86.31-88.90% of the shallower track's covered bins, so 11-14% of the SHALLOWER track's coverage falls where the DEEPER one has none. Depth cannot explain that.

also called H3K27me3 is not scoreable · the channel exists and does not agree · a refusal that is a result

A dataset says it repeated one of its measurements, and it did - but the two repeats do not even agree about which parts of the genome carry a signal. More than a tenth of the shallower repeat's signal sits where the deeper one sees nothing at all, which is not something you can blame on it being shallower.

The same thing, for a scientist

**The channel is real.** Cycle 107's parse of the deposit's 158 sample titles found H3K27me3 in genuine same-condition replicate groups. Verified live here: | condition | rep1 (`.filt.bam.tdf`, Mar 2024) | rep2 (bare `.tdf`, Oct 2024) | |---|---|---| | DMSO | GSM8123369 | GSM8549820 | | SGC0946 5 uM | GSM8123370 | GSM8549822 | | VTP50469 500 nM | GSM8123371 | GSM8549821 | All six were downloaded and binned - none was on disk, contrary to the dispatch brief, whose `q125_bin_K27_*` files are the NTvKO arm (GSM8123423-8123428). **And no floor can be computed.** Genuine track agreement `|A n B| / min(|A|,|B|)`: | condition | \|A\| | \|B\| | \|A n B\| | agreement | |---|---:|---:|---:|---:| | DMSO | 1,103,765 | 849,186 | 738,191 | **86.93%** | | SGC | 1,092,605 | 925,535 | 798,863 | **86.31%** | | VTP | 1,133,060 | 866,052 | 769,908 | **88.90%** | **Read the denominator carefully: it is the SHALLOWER track.** So 11-14% of the shallower track's own covered 2 kb bins carry **zero** coverage in the deeper one. Greater depth adds bins; it does not remove them. A shallow library's covered bins should be very nearly a subset of a deeper one's from the same chromatin, and here one bin in eight is not. **Both gates refuse this, which matters.** It aborts under the inherited universe gate (83.61-85.99%) and under the amended one, so the refusal is not an artefact of this beat's amendment. It is the strongest refusal of any comparison attempted.

Go deeper
What could explain it, and what the deposit lets us test

The two H3K27me3 channels differ in both available nuisances - the .filt.bam marker and a seven-month submission batch - so this is the P_both shape, the same shape that was also refused for H2AK119ub (90.43-91.38%). H3K27me3's agreement is worse still.

Candidate explanations the deposit cannot separate: a different filtering threshold between the two pipelines; a genuinely different chromatin preparation; a different antibody lot. The token antibod occurs zero times in the 158 sample records, so the last cannot be checked at all.

What this is NOT. It is not evidence that either channel is wrong, and it is not a claim about H3K27me3 biology. It is a statement that these two files cannot be used to bound each other's noise.

The consequence for the thread

H3K27me3 keeps a transferred floor, and now for a documented reason rather than for want of a replicate. Any H3K27me3 claim in cycles 100-107 - including the largest effect Q125 reported, the Q1 collapse from 2.0007 to 1.0134 in the PCGF1 knockout - is scored against a floor imported from a different mark, and this beat could not fix that. It is the most important thing this beat set out to do and failed to do.

THE RENORMALISATION BOUND, AND IT APPLIES TO EVERY NUMBER IN THIS ENTRY. Within-sample shares sum to 1 and GSE260742 has NO USABLE SPIKE-IN (Q117 verified zero hits for the entire spike-in vocabulary across all 419,824 characters of its 158 sample records, and no scale factor is deposited), so a purely local change leaks into every other stratum by arithmetic. A FLOOR COMPUTED THIS WAY CANNOT DISTINGUISH 'the bulk moved' from 'the peaks moved and the shares renormalised'. A TIGHTER NUMBER IS NOT THEREBY A CLEANER ONE.

Facts and figures
markH3K27me3
armMLLAF9 Control
absence classmeasured-not-analysed
pairing shapeP_both (marker AND submission batch both differ)
floor computedfalse
brief premise correctedthe dispatch brief stated these rep1 tracks were already binned as q125_bin_K27_*; those files are GSM8123423-8123428, the NTvKO arm
absence class noteFOLDED 2026-09-24: this entry originally carried the non-vocabulary token 'measured-not-usable', which is the truer description and is preserved here; 'measured-not-analysed' is the nearest controlled token.
track agreement PCT
SGC86.3SGC: 86.31DMSO86.9DMSO: 86.93VTP88.9VTP: 88.9
inherited gate PCT
SGC83.6SGC: 83.61DMSO84.1DMSO: 84.11VTP86.0VTP: 85.99
pairs
{
  "DMSO": [
    "GSM8123369",
    "GSM8549820"
  ],
  "SGC": [
    "GSM8123370",
    "GSM8549822"
  ],
  "VTP": [
    "GSM8123371",
    "GSM8549821"
  ]
}
track agreement PCT — the numbers behind the chart
{
  "DMSO": 86.93,
  "SGC": 86.31,
  "VTP": 88.9
}
inherited gate PCT — the numbers behind the chart
{
  "DMSO": 84.11,
  "SGC": 83.61,
  "VTP": 85.99
}
Still open
  1. Do the two H3K27me3 channels disagree about the same 11-14% of bins in all three conditions? If the disagreeing set is shared, it is a pipeline or mappability boundary; if it is condition-specific, it is not.
  2. Is there any H3K27me3 pair anywhere in this thread's deposits that shares a pipeline and a batch? If not, H3K27me3's floor cannot be measured from public data as it stands.
Every source behind this entry (1)
  1. E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    In GSE260742's MLLAF9 Control arm, the three declared H3K27me3 rep1/rep2 pairs (GSM8123369/8549820, GSM8123370/8549822, GSM8123371/8549821) share 86.31-88.90% of the shallower member's non-empty 2 kb bins, so 11-14% of the shallower track's covered bins have zero coverage in the deeper one. All three fail a 95% universe gate in both its inherited and its amended form, and no floor was computed from them.

    What this evidence cannot show

    A 2 kb bin with zero coverage in a TDF is a bin with no reads placed in it by that deposit's own pipeline; it is not independently verified against the BAM, which is not deposited in readable form for these samples. The asymmetry argument (shallow-should-be-a-subset-of-deep) assumes the two libraries sample the same chromatin, which is what is in question. SYSTEM: murine MLL-AF9 (KMT2A::MLLT3) myeloid blood cells, Mus musculus, mm10, one lab, one deposit - NOT the infant B-lineage KMT2A::AFF1 disease this graph is about. n = 3 conditions per arm, one cell system; the spread across conditions is reported so stability is visible and IS NOT A CONFIDENCE INTERVAL. THE RENORMALISATION BOUND, AND IT APPLIES TO EVERY NUMBER IN THIS ENTRY. Within-sample shares sum to 1 and GSE260742 has NO USABLE SPIKE-IN (Q117 verified zero hits for the entire spike-in vocabulary across all 419,824 characters of its 158 sample records, and no scale factor is deposited), so a purely local change leaks into every other stratum by arithmetic. A FLOOR COMPUTED THIS WAY CANNOT DISTINGUISH 'the bulk moved' from 'the peaks moved and the shares renormalised'. A TIGHTER NUMBER IS NOT THEREBY A CLEANER ONE.

    Evidence that argues against this (2)
Connected to