Oncofusion

start / graph / ana-q118-gse127507-ran-a-drosophila-spike-in-measured-it-and-deliberately-did-not-use-it

Analysisestablished confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.Q118-menin-at-the-human-index-fusion

GSE127507 ran a Drosophila spike-in, measured it, and deliberately did not use it - so the VTP-50469 founding dataset carries no recoverable scale factor either, and the depositors say so in the sample record

also called GSE127507 normalisation · no spike-in normalisation in the Krivtsov ChIP-seq deposit

The team that made the key menin-drug dataset did add a fly-DNA yardstick to their experiment, but then decided the yardstick showed nothing interesting and threw the measurement away. That means nobody can compare the raw signal between treated and untreated samples, so every comparison in this re-analysis is a ranking inside one sample rather than a before-and-after amount.

The same thing, for a scientist

**All 26 sample records of GSE127507 (human ChIP-seq subseries of GSE127508, PMID 31821784) carry one verbatim `!Sample_data_processing` line**: *"The 181116 samples also contain Drosophila spike-in chromatin as described; these samples were additionally aligned to dm6 as above. Global differences in fly-human chromatin ratios were found to be small, so no additional spike-in normalization was performed in analysis of these samples."* [E1] This is a stronger and cleaner negative than Q117 obtained for GSE260742, where the whole spike-in vocabulary returned zero hits across 158 records. Here the spike-in **exists**, was **measured**, and was **explicitly not used** - and the measurement behind "found to be small" is nowhere reported as a number, so it cannot be inherited even as an approximation. **"Small" is not a scale factor.** Further, **the batch the sentence names is probably not in this subseries at all.** The string `181116` occurs 52 times in the record dump (twice per sample, in the two processing lines) and in **zero of the 26 sample titles**; the title date stamps are 180110 (x10), 180209, 180402 (x4), 180619 (x3), 180726 (x2), 180919 (x2), 181213 (x2) and 190211 (x2) [E1]. The processing block is shared boilerplate across the SuperSeries. **All 26 deposited processed files are `.deduped.tdf`** - raw `igvtools count` coverage, 75-371 MB, no bigWig, no bedGraph, no peak call, no BAM [E1]. The one series-level processed file is a ROSE TSS-signal table described as *"input-subtracted signal within ROSE-calculated regions"* with **no normalisation statement of any kind** [E2]. **Consequence, binding on every Q118 entry:** within-sample shares and within-sample percentile ranks only.

Go deeper
Why this is worth its own entry rather than a caveat

My dispatch brief told me not to inherit a normalisation claim, because the brief given to Q117 had asserted a ChIP-Rx spike-in for GSE260742 on the strength of a committed graph entry and there was none. The same instruction, applied to a different series, returned a different and more interesting answer: not "no spike-in" but "a spike-in whose result was judged uninformative and discarded."

The two cases need different write-ups. A missing spike-in is a gap in the deposit. A discarded one is a scientific judgement by the depositors that global chromatin occupancy did not change much between conditions - which is itself a (weak, unquantified) piece of evidence, and one that happens to point the same way as this beat's finding that the MLL1-N track's total signal barely moves. It cannot be used as a scale factor, but it should not be reported as an absence either.

What the sample records establish about the whole series, item by item

Counted directly from the 134,259-character SOFT dump of all 26 ^SAMPLE records, fetched 2026-09-23 1:

itemvalue
samples26, all Homo sapiens, all NextSeq 500
alignmentbowtie2 v2.1 --very-sensitive (or STAR 2.6.1b for the 181116 batch) to GRCh37/hg19
duplicatesmarked and removed, Picard v2.9.4
deposited processed files26 of 26 .deduped.tdf
VTP-50469 dose330 nM in every one of the 26 records (the series overall_design says "100-330nM"; the per-sample records do not)
exposure3 days for 24 samples; 4 days for the MOLM13 MENIN pair GSM3635683/84 only
spike-in scale factornone, anywhere
depth normalisation of the deposited signalnone stated

The hg19 claim was not taken on trust: it was re-verified from each downloaded file's own per-chromosome extents (9 bp from hg19, 2,818,723 bp from hg38 on chr1, in all nine files) - see ana-q118-the-genome-build-was-verified-from-nine-files-own-chromosome-extents-and-is-hg19.

Facts and figures
scriptanalyses/q118-menin-at-the-human-index-fusion/PREREGISTRATION.md section 1a
statusrun
result summaryA Drosophila spike-in was run, measured, judged small and deliberately not applied; no scale factor is deposited or recoverable; 26 of 26 processed files are raw igvtools .tdf; 330 nM VTP-50469 in every record; the MOLM13 MENIN pair alone is a 4-day exposure.
seriesGSE127507
superseriesGSE127508
n samples26
genome build metadataGRCh37/hg19
spike in statusrun and deliberately unused; no scale factor deposited
absence classmeasured-not-reported
absence class noteThe gap this token describes is the SCALE FACTOR specifically: the fly/human chromatin ratio was measured (the reads were aligned to dm6) and the number was never written down. This entry carries a second, different gap - that no bigWig or peak call was deposited - which is measured-not-reported as well but is not what the token refers to. It does NOT describe the H3K79me2 arm, which is deposited and simply not analysed here.
inputs
[
  "GSE127507 26 sample SOFT records",
  "GSE127507 series record"
]
How this could bite us back
  • A pharmacodynamic assay for a menin inhibitor built on ChIP signal amplitude cannot be calibrated from this dataset, because the dataset's own scale factor was discarded. Any occupancy-based response criterion has to be defined on a within-sample statistic or it will be reading library depth.
  • This is the founding dataset for VTP-50469 and therefore for revumenib's mechanistic rationale. If a reader takes a browser-track amplitude from it as evidence of how much fusion is left, they are reading an unnormalised igvtools count.
Still open
  1. The fly/human chromatin ratio was computed and discarded. Is it in a supplementary table of PMID 31821784? If so, this dataset becomes the only human menin-inhibitor occupancy series with a recoverable scale factor, and a fold-change analysis of the MLL1-N and MENIN arms becomes possible.
  2. Which samples ARE the 181116 batch? They are not in GSE127507 by title date stamp; GSE127508 has four subseries and the batch should be locatable in one of the other three.
  3. The depositors judged the global fly/human ratio 'small'. Is that compatible with the MENIN track losing ~80% of its share at its own top sites? A global ratio near 1 with a large redistribution within the human genome is exactly what this beat measures, so the two statements are consistent - but nobody has checked that explicitly.
Every source behind this entry (2)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    All 26 GSE127507 sample records state that a Drosophila spike-in was present in one batch, that fly-human chromatin ratios were found to be small, and that no spike-in normalisation was performed; all 26 deposited processed files are raw igvtools .tdf coverage; the dose is 330 nM VTP-50469 in every record and the exposure is 3 days except for the MOLM13 MENIN pair at 4 days.

    What this evidence cannot show

    This is a statement about the deposited metadata, not about what the authors did in unpublished analysis. The inference that the 181116 batch is absent from this subseries rests on title date stamps, and a title date stamp is not guaranteed to be the sequencing-run date the processing note refers to; the conclusion that no scale factor is recoverable does not depend on it.

    Evidence that argues against this (1)
    • The paper's own Methods may name a normalisation the GEO records omit. This entry is bounded by the deposited sample records and the series record; the full text was not re-retrieved in this beat, and PMC7227117 is open access, so a reader can check in one fetch. If a per-sample scale factor is in a supplementary table, every within-sample statistic in this beat remains valid and a fold-change analysis additionally becomes possible. Krivtsov AV et al. 2019
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset

    The single series-level processed file of GSE127507 is a ROSE TSS-associated signal table covering DOT1L, MENIN and MLL1n in all three lines, described as input-subtracted signal within ROSE-calculated regions, with no normalisation statement.

    What this evidence cannot show

    The table covers 18 of the 26 samples; the processing note says it excludes H3K79me2 and it contains no input columns. Input subtraction was performed by the depositors with a cell-line-matched input, which for RS4;11 and ML-2 means the DMSO input was used for both conditions because no VTP input exists for those lines.

    Evidence that argues against this (1)
    • An input-subtracted table is not normalisation-free in effect: subtracting a fixed input from two libraries of different depth distorts the two conditions differently. This is a reason to prefer the genome-wide .tdf re-binning (instrument B) for any quantitative claim, and this beat does so for its headline numbers. NCBI Gene Expression Omnibus 2019
Connected to