The CD34+ oncoprotein-negative sample that calibrates the published N/C threshold is in GSE159608, not GSE252378 - so the 1.75 threshold cannot be re-derived from the deposit that carries the 27 patients it is applied to
The published method sets its cutoff by checking how many false hits it makes in a healthy control sample. That control lives in a different public dataset from the patient samples, so anyone re-analysing the patient data alone has no way to check the cutoff.
The same thing, for a scientist
The paper sets its oncoprotein-site threshold by calibration: *"We set a threshold that produced 3 false-positive sites in the CD34+, oncoprotein-negative control sample"*, arriving at **N/C > 1.75 with a-value < 1e-5** and 1,692 sites across 36 samples [E1]. **That control sample is not in GSE252378.** Bounded by reading all 171 sample records for `CD34`, `normal`, `negative`, `control` and `healthy` across `!Sample_title`, `!Sample_description`, `!Sample_source_name_ch1` and all three `!Sample_characteristics_ch1` lines: **zero hits**, and every one of the 171 records declares a KMT2A translocation [E2]. **It is not missing from the public record, only from this accession.** The paper's Data Availability statement names a second accession and says so explicitly: *"The primary CD34+ HSPC control sample, as well as the four MLLr cell lines, 1 primary-patient ALL, 5 primary-patient AMLs, and 2 primary-patient MPALs were profiled in a previous study"*, deposited as **GSE159608** [E3]. **What this costs a re-analysis that uses only GSE252378** - which is what both this beat and q142 did. The threshold's false-positive rate is a property of the calibration sample, so with that sample in another accession the threshold arrives as a **number without its own control**. Re-applying 1.75 to new intervals, as this beat does on refGene promoters, inherits a calibration performed on different intervals in a different cell type, and nothing in GSE252378 can say by how much that misses.
The substitute control this deposit CAN supply, and why it is harder
GSE252378 carries two different antibodies against each end of KMT2A in all 27 samples - N1 Mouse Millipore 05-764 and N2 Rabbit CST D2M7U; C1 Mouse SantaCruz H10 and C2 Mouse Millipore 05-765. Comparing N1 against N2, or C1 against C2, under the published criterion must call zero sites if the criterion is specific, because no protein species is present in one library and absent from the other. That control holds the biology fixed and varies only the reagent, which a CD34+ sample does not. It is used in ana-q144-two-antibodies-against-the-same-end-of-kmt2a-call-hundreds-of-oncoprotein-target-promoters-under-the-published-criterion.
Actionable, and cheap
GSE159608 is public. Pulling its CD34+ HSPC N/C libraries and running them through this beat's extractor would give the threshold its own calibration on the same intervals, and would convert the same-protein control from a substitute into a cross-check. That is one download and one re-run of analyses/q144-unbounded-magnitude/run.py.
| absence scope | GSE252378, all 171 sample records |
|---|---|
| hits | 0 |
| resolved elsewhere | GSE159608 |
| absence class | measured-not-reported |
| absence class basis | The control sample WAS measured - the paper sets its N/C > 1.75 threshold from it - and was not deposited. This is withholding, not absence. |
tokens searched
[ "CD34", "normal", "negative", "control", "healthy" ]
fields searched
[ "!Sample_title", "!Sample_description", "!Sample_source_name_ch1", "!Sample_characteristics_ch1 (all three lines)" ]
- Does GSE159608's CD34+ HSPC sample carry a complete KMT2A N1/N2/C1/C2 set, and what false-positive rate does the 1.75 / 1e-5 criterion produce on refGene promoters there?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The published oncoprotein-site threshold of N/C > 1.75 with a-value < 1e-5 was set to produce 3 false-positive sites in a CD34+ oncoprotein-negative control sample.
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 text. The calibration is on one negative sample and on the authors' own peak intervals, not on promoters.
Evidence that argues against this (1)
- Calibrating on a single negative sample bounds the false-positive rate only for that sample's depth, chromatin and reagent lot; a criterion tuned to 3 false positives in one library can behave very differently in another, and the paper does not report a second negative. Janssens DH et al. 2024
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
No oncoprotein-negative, CD34+, normal or control sample exists among the 171 records of GSE252378; every record declares a KMT2A translocation.
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
Bounded by a case-insensitive search for CD34, normal, negative, control and healthy over !Sample_title, !Sample_description, !Sample_source_name_ch1 and all three !Sample_characteristics_ch1 lines of all 171 records, plus a check that every description matches the pattern 'contains the KMT2A-<partner> translocation'. A negative sample labelled with none of those five tokens and carrying a fabricated translocation string would be missed, which is implausible but is the bound.
Evidence that argues against this (1)
- This absence is about one accession, not about the public record: the sample exists and is deposited elsewhere, which the next evidence item establishes. Reporting it as 'the control does not exist' would have been wrong. Janssens DH et al. 2021
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
The paper's Data Availability statement places the primary CD34+ HSPC control, four MLLr cell lines, one primary ALL, five primary AMLs and two primary MPALs in GSE159608, a separate accession from GSE252378.
GSE159608 - Automated CUT&RUN profiling of MLL-rearranged leukemia samples and cell lines, including the primary CD34+ HSPC control · Janssens DH, Henikoff S (Fred Hutchinson Cancer Center) · NCBI Gene Expression Omnibus 2021What this evidence cannot show
Quoted from the article's Data Availability section as served by PubMed Central (PMC11522475). The contents of GSE159608 were not independently enumerated in this beat, so the CD34+ sample's library composition there is taken on the paper's word.
Evidence that argues against this (1)
- Because GSE159608's records were not enumerated here, it is possible the CD34+ control lacks a complete N1/N2/C1/C2 set, in which case it would not supply the calibration a re-analysis needs either. That check has not been run. Janssens DH et al. 2024
the accession the calibration sample is absent from
- related to →KMT2A / MLL1 — the gene that breaks
the threshold calibrates the KMT2A N-terminal over C-terminal ratio
the calibration sample is primary CD34+ HSPC
the calibration sample this experiment needs, and where it actually lives
the control whose location that entry establishes