Oncofusion

start / graph / absence-no-cell-state-for-ell

Analysisstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q60-cellstate2

ELL has never been placed in a cell state — its entire leukaemia state literature is its own fusion transcript used as a residual-disease marker

also called ELL bounded cell-state absence

Nobody has ever shown what condition ELL holds a cell in. Every paper that pairs ELL with a cell-state word in leukaemia is using the MLL-ELL fusion as a laboratory test for leftover disease, not studying what ELL does.

The same thing, for a scientist

Bounded query on 2026-09-20: `(ELL OR ELL1 OR ELL2)[tiab] AND <18-term cell-state clause> AND <leukaemia clause>` returns **16**, of which **0** survive a title check as cell-state results [E1]. Controls pass: the gene clause alone returns 923 and the state clause alone 96,743. Twelve of the sixteen are measurable-residual-disease monitoring or fusion-detection method papers in which MLL-ELL is an RT-PCR target; the rest are transplant outcome series and reviews. Per rule 80 one full text was opened before recording the absence and it held [E2]. ELL is also pan-essential in DepMap (93.9% of 1,258 lines depleted), so the dependency screen adds no cell-type information either [E3].

Go deeper
What the full text showed, because the rule-80 step is the part that counts

The most likely place for a hidden ELL cell-state panel is a paper that uses MLL-ELL in transformation assays. PMID 34431785 (HBO1-MLL interaction promotes AF4/ENL/P-TEFb-mediated leukemogenesis, eLife 2021) was opened via PMC8387021: 115,752 characters, seven occurrences of colony-forming and five of self-renewal 2.

And the absence held, because MLL-ELL there is the construct, not the subject. The variable under test is MLL's own trithorax homology domain 2 and its HBO1/KAT7 contact; MLL-ELL is the immortalisation substrate those domains are mapped in. No cell-state function is attributed to ELL anywhere in the paper.

That is a distinction a keyword index cannot make and a title check cannot make, and it is why the absence is recorded with the full text named rather than with a count alone.

The shape of this absence is worth generalising

ELL, MLLT10 and to a large extent AFF1 share it: the gene's name appears throughout the leukaemia literature because its fusion transcript is a diagnostic assay target. The recall is real, the records are on the gene, and none of them is about the gene's function. A count of 16 or 29 reads as a modest but healthy literature and is nearly all diagnostics.

The tell is the journal and article type distribution: FISH panel development, biochip assays, ddPCR, RT-PCR standardisation. If a gene's cell-state hits are concentrated there, the count is measuring how clinically useful the fusion is, not how well the gene is understood.

Facts and figures
genegene-ell
symbolELL
uniprotP55199
statusrun
query(ELL[tiab] OR "ELL1"[tiab] OR "ELL2"[tiab]) AND <cell-state clause> AND <leukaemia clause>
count16
on topic after title check0
on topic after opening full texts0
gene clause alone control923
result summaryELL has no cell-state characterisation in leukaemia or in normal haematopoiesis. Its state-word recall is entirely diagnostic use of the MLL-ELL transcript.
state clause alone control96743
date2026-09-20
toolNCBI esearch db=pubmed
absence classmeasured-not-analysed
absence class basisARCHIVE CHECK RUN 2026-09-21 (king), which is what moved this off unclassified-not-checked. GEO holds 760 human bone-marrow single-cell RNA-seq DataSets (query: "Homo sapiens"[Organism] AND "bone marrow"[All Fields] AND ("single cell" OR "scRNA-seq")[All Fields] AND "expression profiling by high throughput sequencing"[DataSet Type]; translation recorded, ran clean). GSE120221 was verified individually: 25 samples, Homo sapiens, human bone marrow by scRNA-seq with matched mass and flow cytometry. WHAT THAT DATA CAN SETTLE: which haematopoietic populations express ELL and whether its expression tracks a state such as HSC quiescence versus cycling - the associative placement this entry's own literature queries were looking for. WHAT IT CANNOT: function. A functional placement needs perturbation in the state, and that remains never-measured. So the entry is split - the descriptive half is a desk job on deposited data, the functional half still needs an experiment.
absence class assigned2026-09-21 (cycle 47, king)
full texts opened
[
  "34431785 (PMC8387021, 115,752 chars)"
]
cell state clause terms
[
  "quiescence",
  "quiescent",
  "dormancy",
  "dormant",
  "drug-tolerant",
  "drug tolerant",
  "persister",
  "persisters",
  "leukemia stem cell",
  "leukemia stem cells",
  "leukaemia stem cell",
  "leukaemia stem cells",
  "self-renewal",
  "self renewal",
  "differentiation block",
  "differentiation arrest",
  "minimal residual disease",
  "measurable residual disease"
]
archive checked
{
  "archive": "GEO DataSets",
  "date": "2026-09-21",
  "datasets_matching": 760,
  "verified_example": "GSE120221 (25 samples, Homo sapiens, human bone marrow scRNA-seq)",
  "closes_the_descriptive_half": true,
  "closes_the_functional_half": false
}
How this could bite us back
  • ELL is pan-essential at 93.9% of screened lines, so it is not a target; the question is only whether it carries information about the fusion's mechanism.
  • Treating the 16-record count as evidence that ELL is studied would put an unstudied gene into a target rationale.
Still open
  1. Does ELL contribute anything to the KMT2A::ELL phenotype beyond providing an elongation-factor moiety, or is it interchangeable with the other elongation partners?
  2. Would an ELL degron in a KMT2A::ELL line produce a state change that whole-gene CRISPR cannot show?
Every source behind this entry (3)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    A bounded PubMed query pairing ELL and its aliases with an eighteen-term cell-state clause and a leukaemia clause returns 16 records, none of which is a cell-state result on title check; the gene clause alone returns 923 and the state clause alone 96,743.

    NCBI PubMed E-utilities esearch, db=pubmed, queried 2026-09-20 · NCBI Entrez Programming Utilities 2026
    What this evidence cannot show

    Bounded by TITLES AND ABSTRACTS; supporting figures not searched. The alias ELL2 was included deliberately and its hits title-checked, since ELL2 is a different gene. Counts as of 2026-09-20.

    Evidence that argues against this (1)
    • PMID 34431785 does contain colony-forming and self-renewal assays in which MLL-ELL is used, so a reader could reasonably argue that ELL 'appears in' cell-state experiments. The claim here is narrower and the distinction is the point: ELL is the construct, not the variable, and no cell-state function is attributed to it. Takahashi S et al. 2021
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    The full text of PMID 34431785 contains colony-forming and self-renewal assays in which MLL-ELL is the immortalisation construct used to map MLL's own THD2/HBO1 requirement, and attributes no cell-state function to ELL.

    HBO1-MLL interaction promotes AF4/ENL/P-TEFb-mediated leukemogenesis. · Takahashi S, Kanai A, et al. · eLife 2021
    PMID 34431785DOI
    What this evidence cannot show

    One full text of many possible. Read via Europe PMC/PMC8387021 on 2026-09-20, 115,752 characters after tag stripping. A single opened paper bounds the absence better than a count alone but does not bound it completely.

    Evidence that argues against this (1)
    • PUBLISHED CONFLICTING RESULT: a state-relevant function for ELL is published at complex level, in vocabulary no ELL-plus-cell-state query reaches. Luo et al. report that among the SEC family only SEC - the complex containing ELL together with AFF1/AFF4, ENL and AF9 - is required for proper induction of the HSP70 gene upon STRESS, and that AFF4/SEC plays the dominant role in rapid transcriptional induction, with MYC as a direct target whose SEC-dependent expression matters in acute myeloid and lymphoid leukaemia. Stress-inducibility is a cell-state property. Opening one full text (PMID 34431785) bounds this absence better than a count, as the caveat says, but the paper opened uses MLL-ELL as an immortalisation tool and would not contain this content either. Luo Z et al. 2012
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    In DepMap fetched 2026-09-20, ELL is depleted below -0.5 in 93.9% of 1,258 screened lines with a mean gene effect of -1.298.

    What this evidence cannot show

    Floor compression: no dynamic range in which cell-type selectivity could appear.

    Evidence that argues against this (1)
    • AUTHORS' OWN STATED LIMITATION of the instrument behind '93.9% of lines, mean -1.298'. The Chronos authors state at the outset that CRISPR loss-of-function screens 'exhibit a number of biases and artifacts that can confound the results', and Chronos infers gene fitness effects from an explicit model of CELL PROLIFERATION DYNAMICS after knockout. So a pan-essential score is a statement about how quickly a population stops expanding, and floor compression is a compression of proliferation rate specifically. For a gene whose only published leukaemia role is as an elongation factor in a complex that governs stress-response inducibility, a proliferation-rate instrument is not merely saturated - it is measuring a different quantity from the one this entry is looking for. Dempster JM et al. 2021
Connected to