Oncofusion

start / graph / absence-q143-neither-infant-sample-has-an-igg-control-and-neither-does-its-sequencing-batch

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q143-menin-under-drug-in-infants

Neither infant sample has an IgG control, and neither of their two sequencing batches contains one - so every number from this pair is confounded with chromatin accessibility

To know that a protein is really sitting on a gene, you run the same experiment with an antibody that binds nothing and check you get less signal. For these two infant samples that control was never run, and it was not run on anything else processed at the same time either.

The same thing, for a scientist

Of the 171 libraries in GSE252378, twenty-five are IgG. **None of them belongs to 148752 or 152985**, and none was sequenced in either of those samples' batches (`240423` for the menin/ENL/DOT1L CUT&Tag; `240515` for the KMT2A CUT&RUN). The nearest IgG libraries are `R3595_IgG` and `R3652_IgG` from batch `240410` - a **different patient** with a **different fusion** (KMT2A-EPS15), and CUT&RUN only. This corpus has already measured that a 'menin-high' site class is an accessibility class until it is divided by IgG. Every occupancy number from this pair therefore needs an out-of-sample, out-of-fusion reference or none at all, and the choice changes the answer: the raw N-biased-versus-concordant menin contrast in 148752 is **0.74** (menin lower at fusion-biased promoters) and becomes **1.47** once the IgG reference is divided out.

Go deeper
The count, done by hand

25 IgG titles in GSE252378_family.soft 1, at samples p279, p318, p160, p179, p186, p214, p247, p454, A70498, A73051, A80956, A82485, A89691, A93692, A93949, A95941, TB000694, TB040576, TB051939, TB920488, TB971992, TB986292, TB995291, R3595, R3652. The twenty libraries GSM8371001-GSM8371020 - the whole of 148752 and 152985 - contain DOT1L x2, ENL x2, KMT2A-C1, KMT2A-C2, KMT2A-N1, KMT2A-N2 and menin x2 per sample, and no IgG.

What the substitute can and cannot do

R3595_IgG and R3652_IgG are used in this beat as a declared, imperfect accessibility reference, and only for interpretation, never as a divisor inside a headline number. They are:

  • a different patient, so their open chromatin is a different cell's open chromatin;
  • a different fusion (KMT2A-EPS15);
  • CUT&RUN only, so they are not a background control for the menin, ENL and DOT1L channels at all, which are CUT&Tag - a different enzyme, a different tagmentation bias and a different background distribution 3.

Despite that, they are informative: across the 19,404-gene promoter set the IgG libraries correlate with 148752's KMT2A-C channel at Spearman 0.47-0.58, i.e. a large fraction of the promoter signal in the best-behaved channel of the best-behaved sample is shared with a non-specific antibody in a different patient.

Absence class, and what closes it

never-measured. The cells are gone; no reanalysis can produce an IgG library for this patient. What can be done cheaply is to use the deposit's other IgG libraries to build an accessibility prior for primary infant KMT2A-r marrow and subtract that, which is compute and not wet work - so a follow-up analysis is measured-not-analysed even though the control itself is never-measured.

Facts and figures
absence classnever-measured
igg libraries in series25
igg for 1487520
igg for 1529850
igg in batch 240423 cutandtag0
igg in batch 240515 cutandrun0
statusrun
scriptanalyses/menin-under-revumenib-in-an-infant-patient-pair/run2_controls.py
nearest igg
{
  "samples": [
    "R3595",
    "R3652"
  ],
  "batch": "240410",
  "patient": "different",
  "fusion": "KMT2A-EPS15",
  "assay": "CUT&RUN only"
}
measured consequence
{
  "n_biased_vs_concordant_menin_ratio_raw": 0.743,
  "same_after_dividing_by_igg": 1.47,
  "igg_own_ratio": 0.504,
  "note": "the sign of the menin contrast flips on whether an accessibility reference is used"
}
igg vs 148752 KMT2A C promoter spearman
[
  0.47,
  0.58
]
How this could bite us back
  • Any pharmacodynamic claim built on this pair inherits an accessibility confound that cannot be removed retrospectively, and the direction of at least one contrast here depends on whether the confound is addressed.
  • A reader who takes 'menin is undetectable under the drug' from this deposit is reading a within-library share with no background floor; the same data with an IgG track might show menin at background rather than at zero, and those are different clinical statements.
Still open
  1. Can an accessibility prior built from the deposit's other 25 IgG libraries be transferred to these two samples, and does the N-biased menin contrast survive it?
  2. Do CUT&RUN and CUT&Tag IgG libraries from this platform have interchangeable background distributions? The deposit contains no CUT&Tag IgG at all, so this cannot be answered from it.
  3. Would an H3K4me3 or H3K27ac library from these two samples have served as a background reference? Neither exists for this pair, though both exist elsewhere in the deposit.
Every source behind this entry (3)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    GSE252378 contains 25 IgG libraries and none belongs to sample 148752 or 152985 or to either of their sequencing batches.

    What this evidence cannot show

    Counted from `!Sample_title` lines containing 'IgG' in the raw SOFT family file, then cross-checked against the batch code embedded in each title. A library whose title omitted 'IgG' while being an IgG would be missed; none of the 171 titles is ambiguous.

    Evidence that argues against this (1)
    • An IgG control is not the only possible background reference. A histone mark library from the same cells, or the genome-wide 5 kb background of the same channel, also bounds non-specific signal, and this deposit carries H3K4me3 for other samples - just not for these two. Janssens DH et al. 2024
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    This project has previously measured that a menin-high site class is an accessibility class until it is divided by IgG.

    analyses/menin-under-revumenib-in-an-infant-patient-pair - pre-registered re-analysis of GSE252378 samples 148752 and 152985 · Q143-menin-under-drug-in-infants (Oncofusion Knowledge Graph) · this repository 2026
    What this evidence cannot show

    That measurement was made in the SEM cell line on a different deposit (GSE159608), so it establishes the hazard rather than the magnitude here.

    Evidence that argues against this (1)
    • The measurement that established the accessibility confound was made in the SEM cell line on a different deposit with a different antibody panel, so it establishes that the hazard exists in this assay family without bounding its size in primary infant marrow - where the cell mixture, and therefore the accessibility landscape, is quite different from a cell line's. Janssens DH et al. 2018
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The menin, ENL and DOT1L libraries here are CUT&Tag and the KMT2A libraries are CUT&RUN, so a CUT&RUN IgG is not a background control for the CUT&Tag channels.

    CUT&Tag for efficient epigenomic profiling of small samples and single cells · Kaya-Okur HS, Wu SJ, Codomo CA, Pledger ES, Bryson TD, Henikoff JG, Ahmad K, Henikoff S · Nature Communications 10:1930 2019
    PMID 31036827DOI
    What this evidence cannot show

    The two methods differ in whether a protein A-Tn5 fusion tagments in situ or a protein A-MNase cleaves; their background distributions differ accordingly. The specific magnitude of that difference in these libraries was not measured.

    Evidence that argues against this (1)
    • Both protocols in this deposit are the authors' automated platform, run on the same robot with the same antibody-binding step, so their backgrounds may be more alike than the published methods comparison implies. Janssens DH et al. 2018
Connected to