start / graph / ana-q130-in-a-human-line-and-an-independent-deposit-an-aggregate-of-0-998-hides-a-monotone-eviction-gradient

AnalysismoderateQ130-does-the-eviction-gradient-exist-in-h3k79me2

In MOLM13, human KMT2A::MLLT3, from the VTP-50469 founding paper's own series GSE127507, H3K79me2's aggregate response at MLL1-N's top sites is 0.998 - and the same sites split by eviction depth run 0.7746 / 1.0601 / 1.0995 / 1.1421 at Spearman -1.000, with the INPUT stratified identically staying flat

also called the human leg · Q130 DEVIATION 5 · 0.998 is not nothing · the gradient in an independent deposit

In human leukaemia cells from a completely different published dataset, averaging says the menin inhibitor does nothing at all to the activating tag - the number is 0.998 out of 1. Splitting the same sites by how much fusion protein the drug removed shows a clean trend from a 22 percent loss to a 14 percent gain.

The same thing, for a scientist

**A human cell, a different deposit, a different stratifier, and the cleanest demonstration in the beat that an aggregate can be exactly 1 while the underlying structure is monotone** [E1]. `GSE127507` is a SubSeries of `GSE127508`, the deposit of PMID 31821784 - the VTP-50469 founding paper. MOLM13 is human KMT2A::MLLT3 AML with an **endogenous** fusion and no epitope tag. 330 nM VTP-50469 for 3 days against 0.33% DMSO. All tracks were binned by Q120 in an earlier cycle at 2 kb on hg19, chr1-22+X [E1]. **Stratified by MLL1-N eviction depth** (the exposure-matched stratifier, both arms at d3), over the 1,389 bins of MLL1-N's DMSO top 0.1%: | contrast | Q1 most evicted | Q2 | Q3 | Q4 least | Spearman | span | **aggregate** | |---|---:|---:|---:|---:|---:|---:|---:| | **H3K79me2** | **0.7746** | 1.0601 | 1.0995 | **1.1421** | **-1.000** | -0.3675 | **0.998** | | **DOT1L occupancy** | **0.1364** | 0.3083 | 0.3555 | 0.3913 | **-1.000** | -0.2549 | 0.2773 | | **INPUT** (negative control) | 0.9497 | 0.9740 | 0.9603 | 0.9377 | +0.400 | +0.0120 | 0.9542 | **The aggregate is 0.998.** On that number alone the menin inhibitor does not touch H3K79me2 at MLL1's top sites in a human MLL-rearranged line - and the same 1,389 bins, split by how much MLL1-N the drug removed, run monotonically from a 22.5% loss to a 14.2% gain, with **0 of 20 matched-null draws reaching the observed span** [E1]. **Stratified by MENIN eviction depth instead** - the drug's own direct target - the same mark gives 0.6437 / 0.6650 / 0.7061 / 0.9005, Spearman **-1.000**, span 0.2568, aggregate 0.7107. Two different stratifiers in the same cells, same sign [E2]. **And the INPUT stratified identically is flat** (Spearman +0.400, span 0.0120, empirical p = 0.90). That is the negative control the murine arm could not supply, because all three murine inputs were themselves graded [E1].

Go deeper

Five ways this is weaker than the murine test, stated before anyone quotes it

No tagged fusion. MLL1-N is a shared epitope that sees wild-type KMT2A as well as the fusion, and MENIN is the drug's own direct target and is evicted by definition. Neither is S_FLAG.
No third culture. The stratifier and the measurement are the same MOLM13 libraries, so a shared-library artefact is not structurally excluded here as it is in the murine NTvKO arm.
No DOT1L-inhibitor arm. GSE127507 has no SGC0946 sample, so the shape control that makes the murine result interpretable does not exist in this series.
Quartiles by rank, because Q117's murine cut values (0.0992 / 0.1748 / 0.2717) are properties of a mouse FLAG ratio and have no meaning on a human MLL1n ratio.
n = 1 per antibody per condition, and the MENIN pair is a 4-day exposure while everything else is 3 days - Q120's own pre-registered hazard, which is why the MLL1n-stratified result is the headline one here.

And the series ran a Drosophila spike-in, measured it, and deliberately did not use it. No scale factor is deposited, so this arm is a within-sample share like every other number in this beat.

The one download this beat needed, and why it is a finding

C:/Temp/q118 has been deleted. It held q118_hg19_blacklist.bed, q118_hg19_chroms.json and the Q118Bin class. The binned human TSVs survive in C:/Temp/q120, but the universe definition did not. hg19 chromosome sizes were recovered from C:/Temp/q125/q125_hg19_chroms.json (chr1 = 249,250,621, correct for hg19) and the hg19 ENCODE blacklist v2 had to be re-fetched: 7,731 bytes, 834 intervals, of which 831 land on chr1-22+X.

That 7.7 kB is the entire download cost of this beat, and it is reported rather than absorbed because the pattern matters: a beat's binned data was kept and its universe definition was thrown away, which makes the data unusable until someone notices and re-fetches. The same is true of any other scratch directory that has been cleaned.

Facts and figures

script	analyses/q130-does-the-eviction-gradient-exist-in-h3k79me2/deviation5_the_human_independent_series.py
status	run
series	GSE127507, SubSeries of GSE127508, PMID 31821784 (Krivtsov 2019, the VTP-50469 founding paper)
line	MOLM13, human KMT2A::MLLT3 (MLL-AF9) AML
build	hg19, 2 kb, chr1-22+X; chromosome sizes from C:/Temp/q125/q125_hg19_chroms.json (chr1 249250621)
result summary	H3K79me2 aggregate 0.998 over MLL1-N's top 0.1%; stratified 0.7746 / 1.0601 / 1.0995 / 1.1421 at Spearman -1.000, span 0.3675, p = 0.0476; INPUT flat at +0.400, p = 0.90; MENIN-stratified -1.000, span 0.2568.
absence class	measured-not-analysed
absence class note	GSE127507's MOLM13 tracks were binned by Q120 in an earlier cycle and used for a four-way ordering of RETAINED occupancy. The eviction-depth stratification was never applied to them, and the beat's own open question assumed no human series existed with both arms.
absence class assigned	2026-09-23 (Q130)
epigenetic readout	H3K79me2 and DOT1L occupancy share ratios under a menin inhibitor in human MOLM13, stratified by MLL1-N and by MENIN eviction depth

inputs

[
  "GSM3635685/686 MLL1n DMSO/VTP d3",
  "GSM3635683/684 MENIN DMSO/VTP d4",
  "GSM3635681/682 DOT1L DMSO/VTP d3",
  "GSM3635687/688 H3K79me2 DMSO/VTP d3",
  "GSM3635689/690 input DMSO/VTP d3",
  "ENCODE hg19 blacklist v2 (re-fetched)"
]

universe

{
  "grid_bins": 1518163,
  "blacklist_intervals_used": 831,
  "bins_in_universe": 1389438,
  "k_top_0.1pct": 1389,
  "defined_by": "MOLM13_INPUT_DMSO > 0 and not in the hg19 ENCODE blacklist v2"
}

the one download

{
  "what": "ENCODE hg19 blacklist v2, 7,731 bytes gzipped, 834 intervals",
  "why": "C:/Temp/q118 has been DELETED and it held q118_hg19_blacklist.bed, q118_hg19_chroms.json and Q118Bin. The binned human TSVs survive in C:/Temp/q120 but the universe definition did not.",
  "url": "https://raw.githubusercontent.com/Boyle-Lab/Blacklist/master/lists/hg19-blacklist.v2.bed.gz",
  "reported_rather_than_absorbed": true
}

MLL1n stratified

{
  "stratifier": "MLL1n",
  "S_n": 1389,
  "n_per_quartile": [
    347,
    347,
    347,
    348
  ],
  "eviction_ratio_quartile_medians": [
    0.6667,
    0.9185,
    1.078,
    1.2814
  ],
  "quartiles_by": "RANK of the stratifier's own VTP/DMSO share ratio, equal quarters",
  "contrasts": {
    "H3K79me2": {
      "R_by_quartile_Q1_to_Q4": [
        0.7746,
        1.0601,
        1.0995,
        1.1421
      ],
      "spearman_vs_eviction_depth": -1,
      "span_Q1_minus_Q4": -0.3675,
      "R_aggregate_over_S": 0.998,
      "median_per_bin_log2_by_quartile": [
        -0.1019,
        0.116,
        0.159,
        0.215
      ],
      "bootstrap_span_ci95": [
        -0.407,
        -0.3266
      ],
      "matched_null_median_by_quartile": [
        0.9392,
        0.9463,
        0.9442,
        0.9367
      ],
      "null_span_max_abs": 0.053,
      "n_null_draws_reaching_observed_span": 0,
      "empirical_p_span": 0.0476
    },
    "DOT1L": {
      "R_by_quartile_Q1_to_Q4": [
        0.1364,
        0.3083,
        0.3555,
        0.3913
      ],
      "spearman_vs_eviction_depth": -1,
      "span_Q1_minus_Q4": -0.2549,
      "R_aggregate_over_S": 0.2773,
      "median_per_bin_log2_by_quartile": [
        -2.4441,
        -1.7075,
        -1.6069,
        -1.4253
      ],
      "bootstrap_span_ci95": [
        -0.2813,
        -0.2274
      ],
      "matched_null_median_by_quartile": [
        0.4049,
        0.4072,
        0.4091,
        0.4121
      ],
      "null_span_max_abs": 0.0857,
      "n_null_draws_reaching_observed_span": 0,
      "empirical_p_span": 0.0476
    },
    "INPUT": {
      "R_by_quartile_Q1_to_Q4": [
        0.9497,
        0.974,
        0.9603,
        0.9377
      ],
      "spearman_vs_eviction_depth": 0.4,
      "span_Q1_minus_Q4": 0.012,
      "R_aggregate_over_S": 0.9542,
      "median_per_bin_log2_by_quartile": [
        -0.0704,
        -0.0222,
        -0.0519,
        -0.062
      ],
      "bootstrap_span_ci95": [
        -0.0222,
        0.0449
      ],
      "matched_null_median_by_quartile": [
        0.9413,
        0.9311,
        0.9172,
        0.9004
      ],
      "null_span_max_abs": 0.0642,
      "n_null_draws_reaching_observed_span": 18,
      "empirical_p_span": 0.9048
    },
    "stratifier_itself_TAUTOLOGICAL": {
      "R_by_quartile_Q1_to_Q4": [
        0.579,
        0.9158,
        1.0797,
        1.3289
      ],
      "spearman_vs_eviction_depth": -1,
      "span_Q1_minus_Q4": -0.7499,
      "R_aggregate_over_S": 0.9438,
      "median_per_bin_log2_by_quartile": [
        -0.585,
        -0.1226,
        0.1083,
        0.3577
      ],
      "bootstrap_span_ci95": [
        -0.7834,
        -0.7179
      ],
      "matched_null_median_by_quartile": [
        0.9615,
        0.972,
        0.9691,
        0.9701
      ],
      "null_span_max_abs": 0.0865,
      "n_null_draws_reaching_observed_span": 0,
      "empirical_p_span": 0.0476
    }
  }
}

MENIN stratified

{
  "stratifier": "MENIN",
  "S_n": 1389,
  "n_per_quartile": [
    347,
    347,
    347,
    348
  ],
  "eviction_ratio_quartile_medians": [
    0.0761,
    0.1247,
    0.185,
    0.2821
  ],
  "quartiles_by": "RANK of the stratifier's own VTP/DMSO share ratio, equal quarters",
  "contrasts": {
    "H3K79me2": {
      "R_by_quartile_Q1_to_Q4": [
        0.6437,
        0.665,
        0.7061,
        0.9005
      ],
      "spearman_vs_eviction_depth": -1,
      "span_Q1_minus_Q4": -0.2568,
      "R_aggregate_over_S": 0.7107,
      "median_per_bin_log2_by_quartile": [
        -0.8932,
        -0.849,
        -0.4182,
        -0.0356
      ],
      "bootstrap_span_ci95": [
        -0.3052,
        -0.2091
      ],
      "matched_null_median_by_quartile": [
        0.9565,
        0.9466,
        0.9373,
        0.9429
      ],
      "null_span_max_abs": 0.0465,
      "n_null_draws_reaching_observed_span": 0,
      "empirical_p_span": 0.0476
    },
    "DOT1L": {
      "R_by_quartile_Q1_to_Q4": [
        0.1178,
        0.185,
        0.2472,
        0.3559
      ],
      "spearman_vs_eviction_depth": -1,
      "span_Q1_minus_Q4": -0.2381,
      "R_aggregate_over_S": 0.1974,
      "median_per_bin_log2_by_quartile": [
        -3.873,
        -3.2069,
        -2.4654,
        -1.7395
      ],
      "bootstrap_span_ci95": [
        -0.2688,
        -0.207
      ],
      "matched_null_median_by_quartile": [
        0.4228,
        0.4299,
        0.4136,
        0.4225
      ],
      "null_span_max_abs": 0.0426,
      "n_null_draws_reaching_observed_span": 0,
      "empirical_p_span": 0.0476
    },
    "INPUT": {
      "R_by_quartile_Q1_to_Q4": [
        0.9733,
        1.0028,
        0.997,
        0.9637
      ],
      "spearman_vs_eviction_depth": 0.4,
      "span_Q1_minus_Q4": 0.0096,
      "R_aggregate_over_S": 0.9809,
      "median_per_bin_log2_by_quartile": [
        -0.0108,
        -0.0089,
        -0.0186,
        -0.0196
      ],
      "bootstrap_span_ci95": [
        -0.0275,
        0.0444
      ],
      "matched_null_median_by_quartile": [
        0.9866,
        0.9731,
        0.9296,
        0.9003
      ],
      "null_span_max_abs": 0.1121,
      "n_null_draws_reaching_observed_span": 20,
      "empirical_p_span": 1
    },
    "stratifier_itself_TAUTOLOGICAL": {
      "R_by_quartile_Q1_to_Q4": [
        0.0694,
        0.1238,
        0.1846,
        0.3783
      ],
      "spearman_vs_eviction_depth": -1,
      "span_Q1_minus_Q4": -0.3089,
      "R_aggregate_over_S": 0.1602,
      "median_per_bin_log2_by_quartile": [
        -3.7163,
        -3.0039,
        -2.4341,
        -1.8259
      ],
      "bootstrap_span_ci95": [
        -0.3413,
        -0.2793
      ],
      "matched_null_median_by_quartile": [
        0.6435,
        0.6415,
        0.6451,
        0.6475
      ],
      "null_span_max_abs": 0.0528,
      "n_null_draws_reaching_observed_span": 0,
      "empirical_p_span": 0.0476
    }
  }
}

How this could bite us back

An aggregate of 0.998 in the founding paper's own human line is exactly the number a pharmacodynamic assay would return in a fully drug-exposed patient, and it means 'no engagement'. If the gradient is real, every occupancy-averaged H3K79me2 readout of menin inhibition in a human MLL-rearranged cell has been reading a number that cannot see the drug.
The gradient requires a paired fusion-occupancy measurement to define the strata, and in a human cell with an endogenous untagged fusion that measurement is an MLL1-N or MENIN ChIP - both of which cycle 83 showed move under a DOT1L inhibitor too. So the human assay needs three ChIPs and a drug-class control, which is a serious practical objection to using it clinically.

Still open

Does the gradient exist in a KMT2A::AFF1 cell? Every system with this measurement is KMT2A::MLLT3 - murine MLL-AF9 and human MOLM13 - and this graph exists for infant B-lineage KMT2A::AFF1. Q118 measured the same channels in RS4;11 (KMT2A::AFF1) and those tracks may still be reachable, but C:/Temp/q118 has been deleted.
Why is the human aggregate 0.998 and the murine 0.8156? The two differ in species, in set definition (MLL1-N top sites against FLAG top sites), in dose (330 nM d3 against 500 nM 48 h) and in deposit. Nothing here decomposes that.
Did the founding paper's authors see this? The stratification is one line of analysis on their own deposited data; if they ran it and did not report it, this is measured-not-reported rather than measured-not-analysed.

Every source behind this entry (3)

E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
In MOLM13 (human KMT2A::MLLT3 AML) from GSE127507, a SubSeries of the VTP-50469 founding paper's deposit, 330 nM VTP-50469 for 3 days gives an H3K79me2 aggregate share ratio of 0.998 over MLL1-N's DMSO top 0.1% (1,389 bins) while the same bins split by MLL1-N eviction depth run 0.7746 / 1.0601 / 1.0995 / 1.1421 at Spearman -1.000, span 0.3675, with 0 of 20 matched-null draws reaching that span and the INPUT stratified identically flat at Spearman +0.400.
Q130 re-analysis of GSE260742: DEVIATION 5, the human independent-series test. analyses/q130-does-the-eviction-gradient-exist-in-h3k79me2/run.py, deviation1_per_bin_robustness.py, deviation2_two_criteria_i_wrote_badly.py, PREREGISTRATION.md and results.json, run 2026-09-23. Zero downloads: every track was binned by Q124 and read from disk. · Iqbal Lab oncofusion knowledge graph 2026
What this evidence cannot show
SHARED EPITOPE: MLL1-N (Bethyl A300-086A) detects wild-type KMT2A as well as the fusion. NO THIRD CULTURE: stratifier and measurement are the same MOLM13 libraries. NO DOT1L-INHIBITOR ARM in this series. QUARTILES BY RANK, not by Q117's murine cuts. n = 1 per antibody per condition. WITHIN-SAMPLE SHARE ONLY - the series ran a Drosophila spike-in, measured it and deliberately did not use it, and no scale factor is deposited. THE ONE DOWNLOAD: C:/Temp/q118 has been deleted, so the hg19 ENCODE blacklist v2 (7,731 bytes, 834 intervals) was re-fetched; hg19 chromosome sizes came from C:/Temp/q125. SYSTEM: human KMT2A::MLLT3 MYELOID, not the infant B-lineage KMT2A::AFF1 disease this graph is about.
Evidence that argues against this (2)
- Cycle 83 recorded that the MLL1-N channel moves as much under a DOT1L inhibitor as under a menin inhibitor, so a stratifier built on MLL1-N is not a clean measure of menin-driven fusion eviction. Iqbal Lab oncofusion knowledge graph 2026
- A published methods study establishes that STANDARD ChIP-seq normalisation FAILS TO DETECT a genome-wide decrease in a histone mark under an inhibitor of its writer: standard methods see no H3K27me3 reduction under an EZH2 inhibitor, and a substantial reduction appears only with a Drosophila chromatin spike-in. Every number in this entry is exactly the kind of normalisation that paper shows to be blind. Egan B et al. 2016
E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
Stratifying the same MOLM13 bins by MENIN eviction depth instead of MLL1-N gives H3K79me2 0.6437 / 0.6650 / 0.7061 / 0.9005 at Spearman -1.000, span 0.2568, aggregate 0.7107, with the INPUT again flat at Spearman +0.400 and empirical p = 1.0.
Q130 re-analysis of GSE260742: DEVIATION 5, results.MENIN. analyses/q130-does-the-eviction-gradient-exist-in-h3k79me2/run.py, deviation1_per_bin_robustness.py, deviation2_two_criteria_i_wrote_badly.py, PREREGISTRATION.md and results.json, run 2026-09-23. Zero downloads: every track was binned by Q124 and read from disk. · Iqbal Lab oncofusion knowledge graph 2026
What this evidence cannot show
THE MENIN PAIR IS A 4-DAY EXPOSURE (GSM8123683/684 at 330 nM d4) while the H3K79me2 pair is 3 days - Q120's own pre-registered hazard - so the MENIN-derived strata are defined at a different exposure from the measurement they stratify. MENIN occupancy is also the drug's direct target rather than the fusion's, so a MENIN-defined stratum is 'how much menin the drug removed', which is related to but not identical with 'how much fusion the drug removed'.
Evidence that argues against this (1)
- Two stratifiers drawn from the same experiment are not independent tests; if a single library defect depressed drug-arm signal at high-occupancy sites, both stratifiers and the mark would move together. Iqbal Lab oncofusion knowledge graph 2026
E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
GSE127507 is a SubSeries of GSE127508, the deposit accompanying the VTP-50469 founding paper, which established that a menin-MLL inhibitor induces specific chromatin changes and eradicates disease in models of MLL-rearranged leukaemia.
A Menin-MLL Inhibitor Induces Specific Chromatin Changes and Eradicates Disease in Models of MLL-Rearranged Leukemia. · Krivtsov AV, Evans K, Gadrey JY, et al. · Cancer Cell 2019
PMID 31821784 DOI
What this evidence cannot show
Cited for the identity and provenance of the series and the compound, not for any measurement in this entry. The eviction-depth stratification reported here is not in that paper.
Evidence that argues against this (1)
- Re-analysing a founding paper's own deposit and finding structure its authors did not report invites the question of whether the structure is an artefact of the re-analysis. Iqbal Lab oncofusion knowledge graph 2026

Connected to

supports →In murine MLL-AF9, H3K79me2's response to a menin inhibitor runs 0.586 / 0.819 / 0.928 / 1.187 across the four eviction-depth quartiles of the fusion's own top-0.1% sites - Spearman -1.000, span 0.6005, the largest of any mark measured in these bins except H3K27me3 - while the aggregate over the same 1,228 bins is a single unremarkable 0.8156
the human, independent-deposit replication of the gradient
depends on →MEN1 / menin — the adaptor the fusion cannot work without
MENIN occupancy as the second eviction stratifier
related to →DOT1L — the target where excellent chemistry met an indirect interaction
DOT1L occupancy and its mark measured in the same cells
related to →Revumenib (SNDX-5613, Revuforj) — first approved menin–KMT2A interface inhibitor
the class effect, measured with VTP-50469 in the founding paper's own series
← part ofIn human MOLM13, DOT1L's own occupancy under a menin inhibitor runs 0.1364 / 0.3083 / 0.3555 / 0.3913 across MLL1-N eviction-depth quartiles at Spearman -1.000 - the writer is lost in proportion to eviction depth, and its mark follows with a 2.9-fold smaller span
the writer's own profile in the same human cells
← depends onAcross eight murine contrasts and two human stratifications, the H3K79me2 eviction span separates a menin inhibitor from a DOT1L inhibitor 22-fold with no overlap AND detects menin engagement in a human cell whose aggregate reads 0.998 - two properties the aggregate has one of and one of
the sensitivity half comes from the human 0.998 aggregate
← related toRS4;11 - the human KMT2A::AFF1 index-fusion line - has a deposited H3K79me2 pair, a MENIN pair, an MLL1-N pair, a DOT1L pair and an input under the same menin inhibitor in GSE127507. Every track needed to run this beat's test on the index fusion exists. The download is 1.99 GB measured by HEAD request, and nobody has run it.
the same series, the other human line
← supportsA menin inhibitor's chromatin signature is organised by how much fusion it removed from each site, so every mark's response is a GRADIENT in eviction depth - and any readout that averages over a fusion-occupancy set is underpowered by construction, sometimes to exactly 1.000
the human arm, and the 0.998 aggregate that motivates the hypothesis
← related toDownload and bin GSE127507's nine RS4;11 tracks (1.99 GB), rebuild a human binner, and score the eviction gradient in the index fusion KMT2A::AFF1 - the one measurement that would move this finding from KMT2A::MLLT3 to the disease this graph is about
the control set transferred from the MOLM13 arm