GSE252378 holds eight KMT2A::AFF1 oncoprotein panels from two centres, not the three this corpus had used
We thought this public dataset had three infant leukaemia samples to work with. Reading the sample list properly, it has eight — from two different hospitals. The extra five more than doubled what could be measured.
The same thing, for a scientist
Cycle 125 recorded GSE252378's KMT2A::AFF1 content as p279, p318, p454 and the SEM cell line — "two patients plus a line". Reading `!Sample_characteristics_ch1` across all **171** samples in the series matrix finds **eight sample records carrying the full IgG + KMT2A-N1/N2 + KMT2A-C1/C2 panel**: p279, p318, p454, **p179, p247, TB-05-1939, TB-92-0488 and TB-98-6292** [E1]. Since p279 and p318 are one patient in two lineage states, that is **seven independent patients across two centres** — Texas Children's Hospital and St. Jude — rather than two. The five previously unused records have KMT2A-N/C and IgG only; AFF1, DOT1L, ENL and H3K4me3 remain confined to p279, p318 and p454.
The full inventory, read from the series matrix
| sample | characteristics field says | centre | full N/C/IgG panel | activity tracks |
|---|---|---|---|---|
| p279 | infant B-ALL lineage-switching | Texas Children's | yes | yes |
| p318 | infant AML lineage-switching (the switch of p279) | Texas Children's | yes | yes |
| p454 | infant BALL | Texas Children's | yes | yes |
| p179 | infant BALL | Texas Children's | yes | no |
| p247 | pediatric BALL | Texas Children's | yes | no |
| TB-05-1939 | infant Pre-BALL | St. Jude | yes | no |
| TB-92-0488 | infant Pre-BALL | St. Jude | yes | no |
| TB-98-6292 | infant Pre-BALL | St. Jude | yes | no |
| SEM | pediatric B-ALL cell line | — | no, AFF1 only | AFF1 only |
Note p247 is labelled pediatric, not infant. That matters: Symeonidou reports that all five pediatric patients in the Andersson cohort expressed HOXA9 and clustered with iALL-HOXA9 2. p247 turns out to be the second-highest HOXA-occupancy sample in the deposit, which is a small, independent, blind corroboration of that literature claim.
Why the inventory was wrong, and the generalisable part
The three known samples share a GSM block (GSM8371031-56, plus GSM8000492-501) and are the ones with the rich antibody panel; the five missed samples sit in a different GSM range (GSM8006578-8006632) and carry only the core five tracks. An inventory built from the samples that appear in the paper's main figures, or from a contiguous accession block, misses them.
The check that would have caught it costs one command: group all !Sample_characteristics_ch1 values by fusion and count, rather than listing the samples you already know about. Doing that on this series returns 33 records mentioning KMT2A-AFF1.
| series | GSE252378 |
|---|---|
| n samples in series | 171 |
| independent patients | 7 |
kmt2a aff1 records with full panel
[ "p279", "p318", "p454", "p179", "p247", "TB051939", "TB920488", "TB986292" ]
centres
[ "Texas Children's Hospital", "St. Jude Children's Research Hospital" ]
previously used in this corpus
[ "p279", "p318", "p454", "SEM" ]
newly used in q145
[ "p179", "p247", "TB051939", "TB920488", "TB986292" ]
activity tracks limited to
[ "p279", "p318", "p454" ]
- An under-counted public dataset silently caps the power of every beat that uses it, and nothing in the pipeline reports that a deposit was only partly read.
- The deposit also holds 23 KMT2A-ENL and 10 KMT2A-EPS15 records with the same panel. Does the HOXA/IRX1 axis exist in other KMT2A fusions, or is it t(4;11)-specific as Kuhn's title asserts?
- Are p179, p247 and the three St. Jude samples described in the Janssens paper's supplementary tables under different identifiers? Cross-walking them would recover clinical annotation this beat does not have.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
The GSE252378 series matrix lists eight KMT2A::AFF1 sample records carrying the full IgG + KMT2A-N1/N2 + KMT2A-C1/C2 panel, from Texas Children's Hospital and St. Jude, of which five were unused by prior work in this corpus.
GSE252378 - AutoCUT&RUN profiling of MLL-rearranged leukemia samples (hg38 bigWigs and fragment BEDs) · Janssens DH, Ahmad K, Henikoff S · NCBI GEO 2024What this evidence cannot show
Read directly from !Sample_characteristics_ch1 and !Sample_supplementary_file across all 171 samples in the series matrix. The centre and diagnosis labels are the depositors' free text, not a curated ontology, and 'infant Pre-BALL' versus 'infant BALL' may or may not be a meaningful distinction as used.
Evidence that argues against this (1)
- The deposit contains further KMT2A fusions (KMT2A-ENL, -EPS15, -AFDN, -AF10, -AF9, -AFF2) with their own panels; treating the AFF1 set as the whole deposit would be the same error one level up. Janssens DH et al. 2024
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
All pediatric (11-18 year) patients in the Andersson MLL-AF4 cohort expressed HOXA9 and clustered with the iALL-HOXA9 infants.
HOXA9/IRX1 expression pattern defines two subgroups of infant MLL-AF4-driven acute lymphoblastic leukemia · Symeonidou V, Ottersbach K · Exp Hematol 2021What this evidence cannot show
n = 5 pediatric patients in one reanalysed cohort. The observation is used here only as a blind prediction that p247 would be HOXA-high, which it was; a single concordant case is not a test of the claim.
Evidence that argues against this (1)
- If p247's high HOXA occupancy reflects its depth or handling rather than its age group, the corroboration is spurious; its promoter zero-rates (20.1%/17.0%) are mid-range for the set, which argues against that but does not exclude it. Janssens DH et al. 2024
the five extra patients are what made a seven-patient test possible
- related to →KMT2A::AFF1 (MLL-AF4), t(4;11)(q21;q23)
eight KMT2A::AFF1 sample records with full oncoprotein panels
- related to →KMT2A / MLL1 — the gene that breaks
KMT2A N- and C-terminal antibody panels across seven patients