The PAF complex species is mouse - and its only deposited series is Hoxa9-ER, not an MLL fusion at all
also called GSE21299 provenance · PAF complex immortalisation species · Hoxa9-ER deposition
An entry inferred, rather than knew, that a PAF complex result was done in mouse cells. The deposited data confirms mouse - twelve samples from female C57BL/6 mice - but also shows that the deposited experiment is not an MLL fusion experiment at all; it is a HOXA9 time course.
The same thing, for a scientist
`ana-partner-provenance-corrections` carries the caveat *"The abstract does not state the species explicitly, but 'immortalisation' in this assay class and in the sibling entry gene-rtf1's reading is murine progenitor immortalisation"* [E1]. The inference was right and can now be replaced with a fact. GEO series **GSE21299**, linked to PMID 20541477 by elink, holds **12 samples, all *Mus musculus***, `strain: C57BL/6`, `gender: female`, `age: 6-8 week old`, `cell line: murine bone marrow transduced with epitope tagged forms of Hoxa9`, `cell type: Hoxa9-ER cells`, across time points day 0, 3, 4 and 5 in triplicate [E2]. The recovered full text names MLL-AF9 41 times and MLL-ENL 17 times, so the paper's MLL work is in those two fusions [E3]. **But the deposited series is a HOXA9 oestrogen-receptor time course, not an MLL-fusion experiment.** That is a second-order finding worth recording: a deposition can resolve the *species* question the caveat asked while being the wrong experiment to answer the *fusion* question the graph would ask next. An agent that reads `organism: Mus musculus` off GSE21299 and writes *the PAF complex MLL result is murine* has got the right answer for the wrong reason, and would get a wrong answer by the same method on a paper whose deposited and headline experiments diverge further.
What GSE21299 is and is not
| Field | Value |
|---|---|
| Samples | 12 (Hoxa9-ER at days 0, 3, 4, 5, in triplicate) |
| Organism | Mus musculus, 12/12 |
| Strain / sex / age | C57BL/6, female, 6-8 weeks |
| Cell line | murine bone marrow transduced with epitope tagged forms of Hoxa9 |
| Cell type | Hoxa9-ER cells |
| MLL fusion | none deposited |
The paper's MLL-fusion experiments - the ones ana-partner-provenance-corrections is about - are MLL-AF9 and MLL-ENL by full-text mention count and are not in this series 3. So the species answer is secure (the whole paper is a murine progenitor system) and the fusion answer comes from the full text, not the deposition.
The generalisable warning
This beat's method assumes that a study's deposition is a sample of its experiments. Usually it is. Twice in sixteen resolutions it was not, in opposite directions:
- Here, the deposition is a different experiment from the one being cited (Hoxa9-ER rather than MLL fusion), so it answers the species question by accident of the study being uniformly murine.
- In
ana-unstated-system-ahd-inhibitor-resolved, the deposition covers the in vitro arm and omits the in vivo arm entirely, so it answers a question the graph was not asking and leaves the asked one open.
The common correction is to state, in every provenance claim, which experiment the deposited samples belong to - not merely which organism they came from. !Sample_title and !Sample_source_name_ch1 carry that, and both were readable here.
| script | none - GEO accession-display text endpoint plus NCBI efetch db=pmc full text |
|---|---|
| status | run |
| result summary | Species resolved: 12 of 12 deposited samples are Mus musculus, female C57BL/6, 6-8 weeks. But the deposited series is a Hoxa9-ER time course containing no MLL fusion; the paper's fusion work is MLL-AF9 and MLL-ENL by full-text mention count. A deposition can answer the asked question while being the wrong experiment. |
inputs
[ "PMID 20541477", "GSE21299", "PMC2888888" ]
counts — the numbers behind the chart
{
"samples_total": 12,
"samples_mouse": 12,
"samples_human": 0,
"mll_fusion_samples_deposited": 0,
"timepoints": 4,
"replicates": 3
}entries affected
[ "ana-partner-provenance-corrections" ]
import axes
[ "species: mouse", "the deposited experiment is Hoxa9-ER, not an MLL fusion" ]
- A PAF-complex-directed intervention would target a general transcription-elongation machine used by essentially every RNA polymerase II gene, and a murine progenitor immortalisation assay gives no read on the normal-tissue cost of that.
- If the synergy is AHD-mediated, it will be absent in KMT2A::AFF1, and an agent developed from MLL-AF9 and MLL-ENL data would fail in the index disease for a structural reason already visible in the domain architecture.
- Reading provenance from a deposition that is a different experiment is a method failure that produces a confident correct answer this time and would produce a confident wrong one on the next paper.
- Which fusion carries the PAF-complex synergy result the graph cites - MLL-AF9, MLL-ENL, or both? A mention count cannot settle it and the deposition does not contain the experiment.
- Does the PAF complex synergise with KMT2A::AFF1? The one MLL-AF4 mention in this paper's full text is not evidence either way.
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The PAF complex paper reports that the PAF complex synergises with MLL fusion proteins at HOX loci to promote leukemogenesis, and its abstract does not state the species of the immortalisation assays explicitly.
The PAF complex synergizes with MLL fusion proteins at HOX loci to promote leukemogenesis. · Muntean AG, Tan J, Sitwala K, Huang Y, Bronstein J, Connelly JA, Basrur V, Elenitoba-Johnson KS, Hess JL · Cancer Cell 2010What this evidence cannot show
Identifiers taken from the NCBI esummary record fetched on 2026-09-19. Full text was read from NCBI efetch db=pmc for PMC2888888 after Europe PMC fullTextXML returned HTTP 500.
Evidence that argues against this (1)
- The PAF complex reaches the index fusion through a route the MLL-AF9 and MLL-ENL experiments do not test, because those fusions carry an AHD that KMT2A::AFF1 lacks; the species question the entry asked is therefore the less consequential of the two unknowns in this citation. Biswas D et al. 2011
- E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.dataset
GEO series GSE21299, the series elink links to PMID 20541477, contains 12 Mus musculus samples from female 6-8 week old C57BL/6 mice, annotated 'cell type: Hoxa9-ER cells' and 'cell line: murine bone marrow transduced with epitope tagged forms of Hoxa9', across four time points in triplicate.
Expression data from murine cell line transduced with epitope tagged forms of Hoxa9 (GSE21299) · GEO submitters · NCBI Gene Expression Omnibus 2010What this evidence cannot show
Read on 2026-09-19 from the GEO accession-display text endpoint and enumerated sample by sample. Submitter-entered. This series is a HOXA9 time course and contains no MLL-fusion sample, so it establishes the species of the study's cellular system and not the fusion of the experiments the graph cites. elink also returned a second, unnumbered GEO DataSet record for this PMID which was not separately parsed.
Evidence that argues against this (1)
- A deposition that turns out to be a different experiment from the cited one is the clearest argument against reading provenance from depositions alone, and the same beat found the opposite shape - a paper broader than its deposition - twice. Rogic S et al. 2026
- E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The paper's recovered full text names MLL-AF9 41 times and MLL-ENL 17 times, alongside THP-1, HL-60, KOPN8, K562 and MLL-AF4, indicating that its MLL-fusion experiments use MLL-AF9 and MLL-ENL.
The PAF complex synergizes with MLL fusion proteins at HOX loci to promote leukemogenesis. · Muntean AG, Tan J, Sitwala K, Huang Y, Bronstein J, Connelly JA, Basrur V, Elenitoba-Johnson KS, Hess JL · Cancer Cell 2010What this evidence cannot show
Counted by regular-expression tally over 62,985 characters of flattened NCBI efetch db=pmc text on 2026-09-19. A mention count is not an experiment inventory - introduction and discussion mentions are counted identically to figure-bearing ones - so this is graded moderate rather than established and should be replaced by a Methods reading if the fusion identity becomes load-bearing for an entry.
Evidence that argues against this (1)
- MLL-AF4 appears once in the same text, so the paper is not wholly silent on the index fusion; a single mention is far too thin to support any claim about it, which is exactly why a mention count cannot substitute for reading the Methods. Kerry J et al. 2017
replaces an inference about species with a deposited fact, and shows the deposited experiment is Hoxa9-ER not an MLL fusion
a case where the deposition answers the asked question while being the wrong experiment
resolved source in the unstated-system worklist