GSE252378 fragment BEDs: a complete KMT2A N-terminal / C-terminal antibody set in 27 patient samples across six fusion partners, deposited and until now unopened by this corpus
also called GSE252378 N/C arm · AutoCUT&RUN oncoprotein score arm
A public dataset holds raw fragment positions from 27 children's leukaemia samples, each profiled with antibodies against both ends of the KMT2A protein. Because only the fusion keeps the front end, comparing the two gives a per-gene readout of where the cancer-driving fusion protein sits.
The same thing, for a scientist
GSE252378 deposits 171 GSMs. **Twenty-seven patient samples carry a COMPLETE set of four KMT2A libraries** - `KMT2A_N1` (Mouse Millipore 05-764), `KMT2A_N2` (Rabbit CST D2M7U), `KMT2A_C1` (Mouse SantaCruz H10), `KMT2A_C2` (Mouse Millipore 05-765) - and 25 of the 27 also carry an IgG library [E1]. The N antibodies detect the fusion **and** the remaining wild-type allele; the C antibodies detect wild-type only, so the N-over-C ratio per interval is the paper's own oncoprotein score [E2]. **Composition of the 27**, read off `!Sample_description` and `!Sample_characteristics_ch1` for every record, not sampled: KMT2A-AFF1 8, KMT2A-ENL 7, KMT2A-AFDN 4, KMT2A-AF10 3, KMT2A-AF9 3, KMT2A-EPS15 2; four contributing institutions (Texas Children's, St Jude, Fred Hutch, Seattle Children's); B-ALL, AML and one infant T-ALL [E1]. **Two of the 27 carry a menin-inhibitor label** - `148752` (B-ALL) and `152985` (its lineage-switched AML), one patient, KMT2A-ENL - and they are the only samples in the deposit with Menin CUT&Tag libraries [E1]. **That label is a CASE-level label, not an aliquot-level one**, and it is misleading for 148752: the paper states that 148752 is the B-ALL *at diagnosis*, before treatment, and that 152985 is the AML that arose during menin-inhibitor treatment [E3]. So the deposit's drug arm is one treated aliquot with one pre-treatment aliquot of the same patient - which is better than no control and is not a vehicle arm, because the two differ in lineage as well as in drug. **Every number derived from these files is a within-sample share. There is no spike-in anywhere in this deposit**, and the deposited bigWigs are already depth-normalised (`bedtools genomecov` scaled by the hg38 reference length), which is why the fragment BEDs and not the bigWigs are the right input for anything that needs a count [E1].
Why the fragment BEDs and not the bigWigs
The deposited .hg38.bw tracks are, in the submitters' own words, "the fraction of counts at each base pair scaled by the size of the hg38 reference sequence" 1. That is a within-sample normalisation applied before deposition, and it discards the library total. A ratio of two bigWig values is therefore already a ratio of shares - usable - but the number of fragments behind each value is gone, so no Poisson or binomial uncertainty can be attached to it. The *_hg38.bed.gz fragment BEDs restore the count, which is what distinguishes an N/C ratio of 8 supported by 400 fragments from one supported by 3.
Retrieval note that costs an hour if missed
!Sample_supplementary_file_1 is the fragment BED for 64 of the 171 records and the bigWig for the other 107; the slots are ordered by filename, not by file type. Select the supplementary file by its .bed.gz extension, never by slot number. See ana-q144-the-geo-supplementary-file-slots-swap-order-and-the-bed-is-in-slot-two-for-107-of-171-records.
Raw-data withholding
!Series_overall_design carries the line "Raw data for 51 Samples are not provided due to donor privacy concerns" 1. This restricts re-alignment but not this use: the processed hg38 fragment BEDs are public for all 171 records, and all 133 files used here downloaded cleanly.
| accession | GSE252378 |
|---|---|
| n gsm total | 171 |
| n samples complete NC set | 27 |
| n igg | 25 |
| genome build | hg38 |
| spike in | NONE |
| files used | 133 *_hg38.bed.gz fragment BEDs, streamed over HTTPS and counted in flight |
fusion partners — the numbers behind the chart
{
"KMT2A-AFF1": 8,
"KMT2A-ENL": 7,
"KMT2A-AFDN": 4,
"KMT2A-AF10": 3,
"KMT2A-AF9": 3,
"KMT2A-EPS15": 2
}antibodies
{
"KMT2A_N1": "Mouse Millipore 05-764",
"KMT2A_N2": "Rabbit CST D2M7U",
"KMT2A_C1": "Mouse SantaCruz H10",
"KMT2A_C2": "Mouse Millipore 05-765"
}menin inhibitor treated
{
"geo_label_applies_to": [
"148752",
"152985"
],
"aliquot_truth_per_paper": {
"148752": "B-ALL at diagnosis, BEFORE menin inhibitor",
"152985": "AML arising DURING menin inhibitor treatment"
}
}paired one patient two states
[
[
"p279",
"p318"
],
[
"148752",
"152985"
]
]- A menin-inhibitor arm with no vehicle control cannot support a drug-effect claim, and reading one out of it would be exactly the confound this corpus has flagged in its own earlier beats.
- Selecting the supplementary file by slot number rather than extension silently substitutes a depth-normalised bigWig for a fragment BED in 107 of 171 records, producing a value that looks like a count.
- Reading GEO's case-level 'treated with the Menin inhibitor' string as an aliquot-level flag turns the untreated diagnostic sample into a treated one, which inverts the direction of any drug comparison built on it.
- Nine of the 27 samples carry fusion partners with no dedicated occupancy entry in this corpus (EPS15 especially, n=2 here) - what does the oncoprotein score look like at partner-specific loci?
- The deposit's only pre-/post-menin-inhibitor pair differs in lineage as well as in drug exposure. Does the submitting group hold an on-treatment B-ALL aliquot that would separate the two?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
GSE252378 contains 27 patient samples with a complete KMT2A N1/N2/C1/C2 AutoCUT&RUN set across six fusion partners, 25 matching IgG libraries, one lineage-switching patient carrying the deposit's only Menin CUT&Tag libraries, deposited hg38 fragment BEDs for every record, and no spike-in.
GSE252378 - KMT2A oncoproteins induce epigenetic resistance to targeted therapies (171 GSMs, AutoCUT&RUN / AutoCUT&Tag, hg38 fragment BEDs and bigWigs) · Janssens DH, Henikoff S (Fred Hutchinson Cancer Center / Van Andel Institute) · NCBI Gene Expression Omnibus 2024What this evidence cannot show
Composition read from the GEO series matrix (all 171 columns of !Sample_title, !Sample_geo_accession, !Sample_description, !Sample_characteristics_ch1 and both supplementary-file fields), which is column-aligned and therefore not exposed to the record-boundary hazard of GEO's brief/SOFT format. Fusion-partner assignment is the submitters' own annotation and was not independently genotyped here.
Evidence that argues against this (1)
- The deposit's own annotation is not infallible: GSM8006584's !Sample_description names a different antibody from its title, its characteristics field and its own filename, so at least one record's free-text description is wrong. Composition counts here were therefore built from the title and characteristics fields, which agree with each other and with the filenames. Janssens DH et al. 2024
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The source paper calls oncoprotein target sites from the ratio of MLL N-terminal to C-terminal signal, at N/C > 1.75 with a-value < 1e-5, yielding 1,692 oncoprotein-binding sites and 1,092 direct target genes across 36 samples.
MLL oncoprotein levels influence leukemia lineage identities · Janssens DH, Duran M, Otto DJ, Wu W, Xu Y, Kirkey D, Mullighan CG, Yi JS, Meshinchi S, Sarthy JF, Ahmad K, Henikoff S · Nature Communications 15:9341 2024What this evidence cannot show
Read from the paper's Results and Methods text, not from the abstract. The threshold was set to produce 3 false-positive sites in a CD34+ oncoprotein-negative control, i.e. it is calibrated on one negative sample.
Evidence that argues against this (1)
- The N/C ratio is a difference between two antibodies as much as between two protein species: N1 and N2 are a mouse and a rabbit antibody and C1 and C2 are two different mouse antibodies, so an N/C ratio absorbs any systematic efficiency difference between those reagents. The paper pools them, which hides that term rather than removing it. Janssens DH et al. 2024
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The paper states that 148752 is the B-ALL sample at diagnosis and 152985 is the patient-matched AML that arose during menin-inhibitor treatment, so GEO's 'treated with the Menin inhibitor' string on both records is a case-level label rather than an aliquot-level one.
MLL oncoprotein levels influence leukemia lineage identities · Janssens DH, Duran M, Otto DJ, Wu W, Xu Y, Kirkey D, Mullighan CG, Yi JS, Meshinchi S, Sarthy JF, Ahmad K, Henikoff S · Nature Communications 15:9341 2024What this evidence cannot show
Read from the article text served by PubMed Central (PMC11522475): 'while 46 oncoprotein-binding sites were detectable in the B-ALL sample at diagnosis, treatment with the menin inhibitor reduced...' and 'the 152985 sample that lineage switched during treatment with the menin inhibitor'. The exact interval between the 148752 draw and the first dose is not stated.
Evidence that argues against this (1)
- GEO's own !Sample_source_name_ch1 and !Sample_description say 'treated with the Menin inhibitor' for BOTH records, so a reader working only from the deposit would reach the opposite conclusion for 148752 - and the deposit is the authority a re-analysis normally uses. The paper is preferred here only because it distinguishes the two aliquots and GEO's string does not. Janssens DH et al. 2024
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
8 of the 27 complete N/C sample sets are KMT2A::AFF1
- related to →KMT2A::MLLT1 (MLL-ENL), t(11;19)(q23;p13.3)
7 of 27 are KMT2A::ENL, including both menin-inhibitor-treated samples
- related to →KMT2A::AFDN (MLL-AF6), t(6;11)(q27;q23)
4 of 27 are KMT2A::AFDN
- related to →KMT2A::MLLT10 (MLL-AF10), t(10;11)(p12;q23)
3 of 27 are KMT2A::AF10 (MLLT10)
- related to →KMT2A::MLLT3 (MLL-AF9), t(9;11)(p21.3;q23.3)
3 of 27 are KMT2A::AF9 (MLLT3)
2 of 27 are KMT2A::EPS15
- related to →KMT2A / MLL1 — the gene that breaks
N-terminal and C-terminal KMT2A antibodies, two of each, in every sample
same deposit; q142 opened the bigWigs and the AFF1/ENL/DOT1L/H3K4me3 arm, this entry opens the fragment BEDs and the N/C arm
infant B-ALL and infant pre-B-ALL samples from four institutions
pediatric AML and lineage-switched AML samples in the same N/C arm
148752/152985 profiled under a menin inhibitor; the paper names revumenib for this lineage-switching patient
inventory of the deposit's N/C arm
retrieval hazard in the same deposit
record-level labelling error in the same deposit
the deposit on which both the bounded and the unbounded statistic were computed
the accession the calibration sample is absent from
the 27 patients the recalibrated threshold would be applied to
the deposit whose oncoprotein score this hypothesis is about
the deposit whose treatment annotation this corrects
the fragment BEDs this was computed from
the 27-sample N/C arm
the two-antibody-per-arm design that makes this control possible
the 27-sample N/C arm
the 27-sample N/C arm
the fragment BEDs scored
the six-partner cohort
library depths from the deposited fragment BEDs
both paired patients
133 promoter count vectors reused from the q144 extraction
institution parsed from the same deposit's sample descriptions
the deposit whose disease labels were enumerated here