GSE252378 holds the only AFF1 CUT&RUN in primary KMT2A::AFF1 patient material - four libraries across two patients, three disease states and the SEM cell line, all hg38

A public dataset contains the only measurements of where the AFF1 protein sits on the DNA of real infant leukaemia patients who carry the KMT2A-AFF1 fusion. There are fewer independent patients in it than the sample names suggest, because two of the three patient samples come from the same child at two different stages of their illness.

The same thing, for a scientist

**The resource, with every accession read back from the GEO record rather than typed.** GSE252378 (171 GSMs) contains AFF1 AutoCUT&RUN in **SEM** (`GSM8371031`, `GSM8371032`), **p279** (`GSM8371033`, `GSM8371034`), **p318** (`GSM8371041`, `GSM8371042`) and **p454** (`GSM8371049`, `GSM8371050`); DOT1L, ENL and H3K4me3 AutoCUT&Tag in the same three patient samples (`GSM8371035`-`GSM8371040`, `GSM8371043`-`GSM8371048`, `GSM8371051`-`GSM8371056`); and KMT2A N-terminal, C-terminal and IgG AutoCUT&RUN in the same patients (`GSM8000492`-`GSM8000496` p279, `GSM8000497`-`GSM8000501` p318, `GSM8006628`-`GSM8006632` p454) [E1]. Processed `*_hg38.bed.gz` fragment BEDs and `*.hg38.bw` bigWigs are public for every one; hg38 was confirmed from each bigWig's own chromosome table (chr1 = 248,956,422) in this beat, not assumed from the filename [E1]. **n is two patients, not three.** GEO's `!Sample_description` for every p279 library reads "infant BALL lineage-switching sample (**paired with p318**)" and every p318 library reads "infant AML lineage-switching sample (**paired with p279**)"; the paper states "p318, the AML that resulted from the lineage switch of p279" [E1][E2]. p454 is an unrelated infant B-ALL with no lineage switch. So the AFF1 series is **two patients in three disease states**, plus one cell line. **`_1` and `_2` are two ANTIBODIES, not two replicates.** AFF1-1 is mouse monoclonal MyBioSource MBS190886, AFF1-2 is rabbit polyclonal Thermo PA5-77068; KMT2A-N1 is Millipore 05-764 (mouse) and N2 is CST D2M7U (rabbit); KMT2A-C1 is Santa Cruz H-10 and C2 is Millipore clone 9-12 [E1][E2]. DOT1L (CST D4O2T), ENL (CST D9M4B) and H3K4me3 (Active Motif 39159) `R1`/`R2` **are** true replicates of one antibody each [E1]. The published analysis pools the two AF4 antibodies [E2]; this beat carries them separately and that decision changes the answer. **Menin is absent from the AFF1 arm.** The only menin libraries in GSE252378 are in the menin-inhibitor-treated KMT2A::ENL pair `148752`/`152985`. No menin CUT&RUN exists in any KMT2A::AFF1 sample in this deposit.

Go deeper

What 'untreated' does and does not mean here

None of the AFF1-arm libraries has an ex vivo drug exposure: the GEO characteristics for p279, p318 and p454 name only the antibody, the patient id, the fusion, the lineage and the hospital, with no treatment field, while the menin-inhibitor samples in the same deposit say so explicitly in three places each 1. But p318 is a post-chemotherapy relapse AML - the paper says the switch happened in response to ALL-directed chemotherapy 2. So 'untreated' is true of the assay and false of the patient, and any statement about p318 is a statement about relapsed, therapy-exposed disease.

A metadata defect worth naming

GSM8371032 (SEM_AFF1_2) carries group: KMT2A-AFF2 in its !Sample_characteristics_ch1. SEM is a KMT2A::AFF1 line and every other SEM or patient library in the series says KMT2A-AFF1. It is a one-character typographical error in the deposit, not a different fusion; it is recorded here because a symbol-matched filter on group would silently drop one of the two SEM AFF1 libraries 1. The same field also reads pedatric for pediatric.

How the accessions were checked, because a parser broke this last cycle

Two independent reads agreeing: (1) each GSM's suppl/ directory listing, whose bigWig filenames embed the sample title (GSM8371033_p279_AFF1_1.hg38.bw); (2) the series matrix transposed column-wise, which cannot produce the !Sample_title-before-!Sample_geo_accession off-by-one that produced a retraction in cycle 121, because in the matrix format each sample is a column and every field is a row. Note the two reads are not fully independent - GEO builds the title from the filename in many submissions - so this is a consistency check, not an adjudication.

Facts and figures

script	analyses/q142-aff1-in-primary-infants/extract.py
status	run
result summary	41 bigWigs scored over 19,404 hg38 promoters; 2 patients, 3 states, 1 cell line; hg38 verified per file.
n gsm total	171
n gsm used	41
genome build	hg38

inputs

[
  "GSE252378",
  "UCSC refGene hg38"
]

Still open

No menin CUT&RUN exists in any KMT2A::AFF1 sample in this deposit - who will run it, and in which of the two patients?
The fragment BEDs would allow fragment-size partitioning (short fragments as TF footprint, long as nucleosomal); nothing in this beat used them.

Every source behind this entry (2)

E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
GSE252378 contains AFF1 CUT&RUN in SEM and three patient samples from two patients, with KMT2A N/C and IgG controls, all hg38, and the two AFF1 libraries per sample use two different antibodies.
GSE252378 - AutoCUT&RUN and AutoCUT&Tag profiles of 36 MLL-rearranged leukemia samples (171 GSMs, hg38) · Janssens DH, Henikoff S (Fred Hutchinson Cancer Center / Van Andel Institute) · NCBI Gene Expression Omnibus 2024
What this evidence cannot show
Accessions and antibodies read from the GEO series matrix and per-sample supplementary listings this cycle. Raw reads are withheld for donor privacy on a subset of the 171 samples; the processed bigWigs and fragment BEDs used here are public for every sample named.
Evidence that argues against this (1)
- GEO metadata can be wrong about its own samples - this very deposit mislabels GSM8371032 as KMT2A-AFF2 - so an accession table read from GEO is a claim about the submission, not about the tube. Nothing here re-genotypes the material. Janssens DH et al. 2024
E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
p318 is the AML that arose from the lineage switch of p279 under ALL-directed chemotherapy, and the published analysis pools the two AF4 antibodies.
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 2024
PMID 39472576 DOI
What this evidence cannot show
Full text read from PMC11522475. Gene symbols are stripped from the PMC plain-text rendering, so symbol-level claims were taken from the GEO record rather than from the running text.
Evidence that argues against this (1)
- The paper describes three lineage-switching patients; pairing statements for the other two pairs (148752/152985, R3595/R3652) were read from GEO descriptions, not confirmed against clinical records, and a deposit's pairing note is not a chart review. Janssens DH et al. 2024

Connected to

related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
AFF1, KMT2A-N/C, DOT1L, ENL, H3K4me3 CUT&RUN in primary KMT2A::AFF1 infants
related to →AFF1 / AF4 — the 3' half of the fusion, and a target with almost no structural footing
two AFF1 antibodies, four samples
related to →SEM (CVCL_0095) — the workhorse, from a five-year-old, and microsatellite unstable
GSM8371031/32 AFF1 AutoCUT&RUN in SEM
related to →Infant acute lymphoblastic leukaemia (ALL diagnosed under 1 year of age)
p279 and p454 are infant B-ALL; p318 is the lineage-switched AML of p279
supports →The infant menin-inhibitor occupancy experiment has been done and deposited - menin, KMT2A and ENL CUT&RUN under revumenib in ONE infant patient sampled before and after a lineage switch, not two patients - but it is KMT2A-ENL, so the index-fusion absence survives, and the same deposit holds untreated KMT2A::AFF1 libraries this corpus cites once
same deposit, now read at sample level
← depends onAFF1 occupancy does not distinguish fusion-occupied promoters from wild-type KMT2A promoters in primary infant KMT2A::AFF1 - reproduced at promoter resolution, and the corpus should stop reading a partner ChIP as a fusion readout
source deposit
← depends onThe taspase-1 N/C antibody control separates fusion-occupied from wild-type KMT2A promoters in every primary infant sample, and it is the C-terminal arm that does the work while being the less reproducible arm
source deposit
← depends onHOXA9/HOXA10 is unoccupied and unmethylated at H3K4 in two of the three primary infant KMT2A::AFF1 states, while the SEM cell line carries the strongest AFF1 signal over that locus of any sample in the deposit
source deposit
← depends onThe B-ALL and the AML of one infant share no more fusion-occupied promoters than two unrelated infants do - lineage, not patient identity, sets where KMT2A::AFF1 sits
source deposit
← depends onAFF1 and fusion occupancy transfer between infants at Spearman 0.82-0.92, but H3K4me3 transfers better at 0.93-0.97 - so the transferable part is mostly promoter structure, not fusion targeting
source deposit
← depends onThe pre-registered magnitude test failed because the statistic was bounded by construction - top-decile share spreads only 1.1-1.4x between samples and cannot express what the published oncoprotein score says varies
no spike-in exists in this deposit
← depends onGSE252378 contains menin CUT&RUN in exactly one patient - a KMT2A::ENL lineage-switching pair under a menin inhibitor - and none in any of its eight KMT2A::AFF1 patients
171 sample records read in full
← related toA second sample id is not a second patient: three of the interesting pairs in GSE252378 are one patient twice, and reading them as independent doubles the apparent n
three paired lineage-switching patients, six sample ids
← related toGSE252378 labels one of its two SEM AFF1 libraries KMT2A-AFF2, a fusion that does not exist in the deposit and does not exist in SEM - a filter on that field silently halves the SEM data
metadata defect in the source deposit
← related toGSE252378 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
same deposit; q142 opened the bigWigs and the AFF1/ENL/DOT1L/H3K4me3 arm, this entry opens the fragment BEDs and the N/C arm
← related toThe dispatch brief scoped this deposit's oncoprotein-score arm at two patients and a cell line; it is 27 patient samples across six fusion partners, and the scoping error was inherited from a beat that only needed three of them
corrects the scope carried forward from the AFF1-antibody arm