Oncofusion

start / graph / ana-q130-the-rep2-samples-geo-title-says-replicate-and-their-filename-says-a-different-culture

Analysismoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q130-does-the-eviction-gradient-exist-in-h3k79me2disputed

GSM8123386/387/388 carry the GEO title 'MLLAF9 Control - ... H3K79me2 rep2' and the supplementary filename 'ChIP_K79_F3_NT_*'. Two committed analyses treat them as a separate culture and one asserts 'FOUR INDEPENDENT MURINE CULTURES'. If the titles are right there are three, one of them replicated. REPORTED AS A DISPUTE, NOT APPLIED.

also called the rep2 / F3_NT discrepancy · is it a replicate or a culture · four independent cultures or three

Three deposited samples are labelled as repeats of an existing experiment in one field and as a different experiment in another. Two earlier analyses followed the filename; the dataset's own title field says repeat. The difference changes how independent a published comparison is.

The same thing, for a scientist

**The discrepancy, from the samples' own records** [E1]: | field | GSM8123386 | |---|---| | `!Sample_title` | *MLLAF9 Control - DMSO H3K79me2* **rep2** | | `!Sample_supplementary_file_1` | `GSM8123386_ChIP_K79_`**`F3_NT`**`_DMSO.tdf` | | `cell line` | MLLAF9 | | `tissue` | Blood | | `cell type` | Myeloid | | `treatment` | 72hrs - identical to GSM8123372 | The same pattern holds for GSM8123387 (SGC) and GSM8123388 (VTP, 48hrs, identical to GSM8123374). **The title says a replicate of the MLLAF9 Control culture; the filename says a non-targeting-guide culture** [E1]. **What follows the filename in this corpus** [E2]: `analyses/q122-h3k79me2-three-arms-with-replicates/results.json` names them `F3NT`, and `analyses/q124-the-timecourse-and-the-pcgf1-knockout/results.json`, field `RESULT_the_floor_moves_in_opposite_directions_under_the_two_drugs`, lists *'F3_NT (Q122, GSM8123386/387/388)'* as one of four cultures and states **'IN FOUR INDEPENDENT MURINE CULTURES THE TWO DRUGS MOVE THE MARK'S MID-OCCUPANCY FLOOR IN OPPOSITE DIRECTIONS, WITH NO OVERLAP.'** **What should change if the titles are authoritative:** three cultures, one of them replicated - MLLAF9 Control (x2), NTvKO Control, NTvKO PCGF1-KO - not four independent ones. **The no-overlap observation itself is unaffected** (the four floor values are 0.7804, 0.6947, 0.7230, 0.7620 under VTP and 1.9918, 2.0094, 1.9478, 1.8905 under SGC, and they still do not overlap). **The words INDEPENDENT and the count FOUR are what change** [E2]. **This is reported as a DISPUTE and not applied, and no file outside this beat was edited.** A GEO title and a submitter's filename disagree and neither is definitive. The argument for the title: *rep2* in a title normally means a replicate, the metadata is otherwise identical, and the series already contains a **separate** arm explicitly labelled NTvKO with its own non-targeting control, which makes a second, differently-named NT culture redundant. The argument for the filename: submitters name files after what they ran. **This entry cannot settle it and does not pretend to** [E1].

Go deeper
Why it matters beyond a word count

For Q124's floor result: 'four independent cultures with no overlap' is a stronger statement than 'three cultures, one replicated, with no overlap'. The observation survives; its weight changes.

For this beat: the same three samples are the replicate pair that DEVIATION 4 uses to estimate span reproducibility. Under the title reading, the 1.99-fold span difference is technical variation; under the filename reading it is between-culture variation, which is larger and makes quoting 1.99 as an error bar conservative. Either way the sign result - four of four at Spearman -1.000 - is unaffected, which is why this beat reported the measurement before resolving the label.

And the general form: a GEO sample carries several fields that can name its identity, and they can disagree. This corpus has repeatedly been bitten by an instrument supplying an answer the data did not; here two fields of one record supply two answers and a beat picked one without recording that a choice was made.

What would settle it

Three routes, in increasing cost: the SRA run metadata for these accessions may carry a library or biosample identifier that groups them with GSM8123372-374 or separates them; the paper's own methods and figure legends (PMID 41634341) may name the replicate structure; and the corresponding author could be asked. None was attempted in this beat, which scoped itself to what was already on disk plus one 7.7 kB blacklist.

Facts and figures
statusDISPUTE - reported, not applied, nothing outside this beat edited
absence classmeasured-not-curated
absence class noteThe GEO title field has been in the stored 158-sample record since Q117 fetched it in cycle 68. Two beats read the same record and used the filename. The discrepancy is not a gap in the data; it is a gap in what was recorded about a choice.
absence class assigned2026-09-23 (Q130)
dispute
{
  "what": "GSM8123386/387/388 carry !Sample_title = 'MLLAF9 Control - ... H3K79me2 rep2' while their supplementary filenames read ChIP_K79_F3_NT_*. The title says a replicate of the MLLAF9 Control culture; the filename says a non-targeting-guide culture.",
  "who_is_affected": "analyses/q122-h3k79me2-three-arms-with-replicates/results.json names them F3NT, and analyses/q124-the-timecourse-and-the-pcgf1-knockout/results.json, field RESULT_the_floor_moves_in_opposite_directions_under_the_two_drugs, lists 'F3_NT (Q122, GSM8123386/387/388)' as one of four cultures and states 'IN FOUR INDEPENDENT MURINE CULTURES THE TWO DRUGS MOVE THE MARK'S MID-OCCUPANCY FLOOR IN OPPOSITE DIRECTIONS, WITH NO OVERLAP.'",
  "what_it_should_say_if_the_titles_are_authoritative": "three cultures, one of them replicated - MLLAF9 Control (x2), NTvKO Control and NTvKO PCGF1-KO - not four independent ones. The no-overlap observation itself is unaffected; the word INDEPENDENT and the count FOUR are what change.",
  "source_that_settles_it": "the samples' own GEO SOFT records, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM8123386 - read this session from the stored 158-sample record and confirmed to carry the identical cell line, tissue, cell type and exposure as GSM8123372.",
  "confidence": "MODERATE, NOT SETTLED. A GEO title and a submitter's filename disagree and neither is definitive; 'rep2' in a title normally means a replicate, and 'F3_NT' in a filename normally means a non-targeting guide. The series contains a SEPARATE arm explicitly labelled NTvKO with its own non-targeting control, which argues the filename's NT is stale rather than meaningful - but that is an inference. REPORTED AS A DISPUTE FOR THE COORDINATOR, NOT APPLIED, AND NO FILE OUTSIDE THIS BEAT WAS EDITED.",
  "consequence_for_this_deviation": "it cuts both ways and is stated in both directions: if they are replicates, this is a reproducibility estimate; if they are two cultures, it is a second-culture replication. EITHER READING SUPPORTS THE GRADIENT, which is why the deviation is reported before the discrepancy is resolved."
}
affected files
[
  "analyses/q124-the-timecourse-and-the-pcgf1-knockout/results.json, field RESULT_the_floor_moves_in_opposite_directions_under_the_two_drugs",
  "analyses/q122-h3k79me2-three-arms-with-replicates/results.json, the F3NT arm naming throughout"
]
accessions
[
  "GSM8123372",
  "GSM8123373",
  "GSM8123374",
  "GSM8123386",
  "GSM8123387",
  "GSM8123388"
]
Still open
  1. Does the SRA run metadata for GSM8123386-388 group them with GSM8123372-374? One elink and one esummary would probably settle it and neither was run in this beat.
  2. Does PMID 41634341's methods section name the replicate structure? The paper is open access and was not consulted for this specific question.
  3. Are there other samples in this 158-record series whose title and filename disagree? Nobody has swept it, and the sweep is one script.
Every source behind this entry (2)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    GSM8123386/387/388 carry '!Sample_title = MLLAF9 Control - ... H3K79me2 rep2' with cell line MLLAF9, tissue Blood, cell type Myeloid and exposures identical to GSM8123372/373/374, while their supplementary filenames read 'GSM8123386_ChIP_K79_F3_NT_DMSO.tdf' and equivalents - the title naming a replicate and the filename naming a different culture.

    What this evidence cannot show

    Read from the stored 158-sample SOFT record this session, the same file Q117, Q122, Q124 and Q125 used and byte-identical across their copies. Neither field is authoritative over the other and this entry does not claim one is. The rep2 TDF's own internal track name reads 'ChIP_K79_F3_NT_VTP.filt.bam', i.e. it agrees with the filename, which is weak evidence for the filename reading - but a file's internal name is written by the same pipeline that named the file.

    Evidence that argues against this (1)
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset

    Two committed analyses in this corpus treat GSM8123386/387/388 as a distinct culture: Q122's results.json names them F3NT, and Q124's results.json field RESULT_the_floor_moves_in_opposite_directions_under_the_two_drugs lists them as one of four and asserts 'IN FOUR INDEPENDENT MURINE CULTURES ... WITH NO OVERLAP', a claim whose count and independence would both change under the title reading.

    What this evidence cannot show

    Read from the committed JSON in code. NO FILE OUTSIDE THIS BEAT WAS EDITED. The underlying floor measurements are not in question - 0.7804 / 0.6947 / 0.7230 / 0.7620 under VTP and 1.9918 / 2.0094 / 1.9478 / 1.8905 under SGC, still non-overlapping - only the description of how independent the four are.

    Evidence that argues against this (1)
Connected to