Oncofusion

start / graph / gene-erg

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q63-hoxa-cofactors

ERG - a heptad stem-cell transcription factor whose 15 mentions in this graph are mostly about OTHER diseases, and whose best-established role in B-ALL is loss of function, not cooperation

also called erg-3 · p55 · ETS-related gene · Transcriptional regulator ERG · Transforming protein ERG

ERG is one of seven transcription factors that work together to keep blood stem cells self-renewing, and too much of it predicts a poor outcome in adult myeloid leukaemia. In childhood B-cell leukaemia the best-established story runs the other way: one subtype works by destroying ERG's function, not by using it.

The same thing, for a scientist

ERG (P11308, 479 aa, 21q22.2) is an ETS-family transcription factor and a member of the **heptad** - SCL/TAL1, LYL1, LMO2, GATA2, RUNX1, ERG and FLI-1 - the combinatorial regulator set defined by genome-wide binding in blood stem/progenitor cells, within which RUNX1, GATA2, SCL and ERG make direct protein-protein contacts that stabilise complex binding to DNA [E1]. Erg is required for HSC self-renewal under transplantation and myelotoxic stress in mice, though heterozygosity suffices at steady state [E2]. In AML, ERG expression is a poor-prognosis risk factor and is driven by the ERG promoters together with a **+85 stem-cell enhancer** and the heptad, with heptad-derived signatures predicting outcome where ERG expression alone fails [E3]. It is also the principal effector of a different leukaemia: EVI1 occupies a conserved intragenic ERG enhancer, and ectopic ERG abolishes the EVI1 dependence of EVI1-driven AML [E4]. **Three things argue against ERG as a KMT2A-fusion cofactor.** First, DepMap: ERG is *less* required in KMT2A-rearranged lines than in other leukaemia lines (z +0.48 myeloid, -0.38 B-lineage; genome-scan rank 15,074 of 17,728, difference +0.054) [E5]. Second, a title-check of `ERG AND (MLL OR KMT2A) AND leukemia` (63 records) found no paper on ERG as a KMT2A-fusion cofactor - the hits are fusion-detection panels, FUS::ERG AML, EWS-ERG Ewing sarcoma and a prostate paper [E6]. Third, and most sharply, in B-ALL the best-characterised ERG biology is **loss of function**: DUX4-rearranged B-ALL, up to 7% of cases, deregulates ERG by frequent deletion or by inducing the dominant-negative isoform ERGalt [E7]. **The graph's 15 ERG mentions were audited for this entry and are dominated by these other diseases, not by a KMT2A axis.**

What it looks like
ERG479 aa
PNT (pointed / SAM domain)PNT (pointed / SAM domain) · residues 113–199ETS DNA-binding domainETS DNA-binding domain · residues 311–3910100200300400
loading AF-P11308-F1…
AlphaFold AF-P11308-F1 ↗
colourshape

This is a prediction, not a measurement. No one has seen this protein in this shape. The colours say how much the software trusts itself, and that is worth reading before anything else — the orange stretches are places the model is effectively saying this part has no fixed shape.

very high (90+)confident (70–90)low (50–70)very low (<50)
I have never read one of these pictures before

A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.

Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.

Anything in yellow sticks is a bound small molecule, which is usually the drug.

Go deeper
What the graph's ERG mentions are actually about - an audit, because the dispatch premise did not survive it

A word-boundary grep of data/graph/nodes on 2026-09-20 returns 15 node files naming ERG. Reading them, the registers are:

registerwhat it is
EWS-ERGa fusion in Ewing sarcoma, a prognostic-subtype analysis
FUS::ERGa rare AML fusion, 1% of a paediatric AMKL cohort; also a transplant-outcome entry
DUX4/ERGaltERG loss of function as the driving event in a B-ALL subtype
ETO2-GLIS2the fusion upregulates and interacts with ERG at enhancers
EVI1EVI1 promoting chemoresistance via ERG
ontogenyfetal-versus-adult permissiveness attributed to ERG, SPI1, GATA1 and CEBPA activity

Not one of them is about ERG as a factor the KMT2A fusion drives or depends on. ERG's presence in this graph is real and its axis is not the one the dispatch assumed. It belongs to the comparative-fusion layer and to the heptad/stem-cell layer.

A token hazard on top: hERG (KCNH2) appears in two entries of this graph in its cardiac-liability sense, next to the QTc discussion on the menin inhibitors. A case-insensitive or substring count of 'ERG' conflates a transcription factor with an ion channel that is a drug-safety endpoint in the same disease.

Expressed in / required in / maintains a state
  • Expressed in. Blood stem/progenitor cells, where it is one of ten regulators profiled genome-wide and one of seven in the heptad 1; AML blasts, where it is part of normal and leukaemic stem-cell signatures and high expression is a poor-prognosis risk factor 3; EVI1-driven AML, aberrantly 4. In KMT2A-rearranged material specifically, no expression measurement was found this cycle.
  • Required in. Murine HSC self-renewal under stress - transplantation or myelotoxic recovery - but not at steady state, where heterozygosity suffices 2. EVI1-driven AML, selectively 4. Not required in KMT2A-rearranged lines; if anything the reverse - DepMap gives KMT2A-rearranged myeloid lines z +0.48 against other leukaemia lines, and the whole-genome scan puts ERG in the 85th percentile in the anti-selective direction 5.
  • Maintains a state. The HSC self-renewal state under haematopoietic stress 2 and, as a heptad member, the stem-cell programme whose activity in AML predicts outcome 3. In EVI1-driven AML, ERG maintains the undifferentiated state: suppressing it induces terminal differentiation 4.
The direction of the B-lineage evidence is backwards from the premise

In DUX4-rearranged B-ALL - a subtype comprising up to 7% of B-ALL - DUX4 rearrangement and overexpression is present in all cases and is accompanied by transcriptional deregulation of ERG, expression of the novel isoform ERGalt whose non-canonical first exon is transcribed from a DUX4 binding site, and frequent ERG deletion. ERGalt retains ERG's DNA-binding and transactivation domains but inhibits wild-type ERG activity and is itself transforming 7. The authors describe it as a unique paradigm in which a transcription factor's deregulation results in loss of ERG function, by either of two routes.

That two independent mechanisms converge on the same endpoint is strong genetic evidence that ERG loss is required in that subtype. It is also the only well-powered, B-lineage, paediatric statement about ERG in acute lymphoblastic leukaemia, and it says ERG is something a B-ALL driver gets rid of.

This does not refute an ERG role in KMT2A-rearranged B-ALL - different driver, different subtype - but it is the reason this entry does not treat ERG as a presumptive cooperating factor in B-lineage disease. The prior from the neighbouring subtype points the other way.

Identifier hazards - ERG is the worst three-letter token in this gene set
  1. gene_exact:ERG AND organism_id:9606 AND reviewed:true returns TWO reviewed human records (verified live 2026-09-20): P11308 ERG, 479 aa, Transcriptional regulator ERG; and Q12809 KCNH2, 1,159 aa, which carries ERG, ERG1 and HERG as gene synonyms. Rule 62 live. The 680 aa gap means verify-accessions.py catches a swap, but a script taking the first data row does not.
  2. The KCNH2 collision is not academic in this graph. hERG is the cardiac potassium channel whose blockade causes QT prolongation, and QTc is a listed adverse event of the menin inhibitors this project is built around. So 'ERG' in this corpus can mean a transcription factor or a cardiac safety liability, and the two appear in the same document set.
  3. ERG is three letters that occur inside ordinary words (energy, emergence, allergen, Berg) and inside surnames and other identifiers. A substring count is meaningless and even a word-boundary count needs title-checking. Measured: ERG[tiab] returns 13,832 records, the overwhelming majority of which are prostate cancer (TMPRSS2-ERG) and cardiac electrophysiology, plus electroretinography, for which ERG is also the standard abbreviation.
  4. Distinguish from the closely related ETS factor FLI1, with which ERG shares the heptad, an Ewing-sarcoma fusion role and high ETS-domain identity.
Facts and figures
uniprotP11308
gene symbolERG
length AA479
alphafold IDAF-P11308-F1
ensemblENSG00000157554
hgncHGNC:3446
cytoband21q22.2
protein classETS-family winged-helix-turn-helix transcription factor, ERG/FLI1 subfamily
resolution notegene_exact:ERG AND organism_id:9606 AND reviewed:true returned TWO reviewed rows, fetched live 2026-09-20 with gene_primary and length requested. Only P11308 has gene_primary ERG (ERG_HUMAN, 479 aa, no gene synonyms). The second row is Q12809 KCNH2 (1,159 aa), which carries ERG, ERG1 and HERG as SYNONYMS. Rule 62 live; the row whose primary name matched was kept and the other discarded. Ensembl id from HGNC's single primary ensembl_gene_id (HGNC:3446), per rule 78.
domains noteRead back from the UniProt flat file for P11308 on 2026-09-20 (FT DOMAIN 113..199 PNT; FT DNA_BIND 311..391 ETS). The transactivation regions are annotated as disordered in the record and no boundary is asserted for them.
cell typemurine blood stem/progenitor cells (heptad, E1); murine adult HSC under transplantation or myelotoxic stress (E2); adult human AML blasts (E3); human and murine EVI1-driven AML (E4); human paediatric/AYA DUX4-rearranged B-ALL blasts (E7). NOT measured in any KMT2A-rearranged cell.
cell statethe stressed, cycling, self-renewing HSC state (E2) and the undifferentiated AML stem-cell state (E3, E4). In DUX4-rearranged B-ALL the relevant state is created by REMOVING ERG function (E7).
subcellular localizationNucleus (UniProt P11308, PROSITE-ProRule inference). The record additionally notes ERG localised in cytoplasmic mRNP granules containing untranslated mRNAs, from an IMP1 ribonucleoprotein-granule proteomics study (PubMed:17289661) - that paper was NOT read this cycle and is recorded as a database statement, not a read result. A cytoplasmic pool for a transcription factor is unusual enough to be worth someone checking.
subnuclear localizationNOT ESTABLISHED for ERG in any leukaemia context by anything read this cycle. Chromatin occupancy is extensively mapped (E1, E4) but occupancy is not localisation.
named but unwritten count at dispatch19
named but unwritten count measured15
what this entry does not claimThat ERG is a cooperating transcription factor downstream of a KMT2A fusion. No evidence for that was found, the dependency data point the other way, and the best-powered B-lineage evidence is about ERG loss of function in a different subtype.
collision
{
  "class": "two_reviewed_human_records_for_one_exact_symbol_plus_a_safety_endpoint_register",
  "other_record": {
    "accession": "Q12809",
    "symbol": "KCNH2",
    "length_aa": 1159,
    "why_it_matches": "carries ERG, ERG1 and HERG as gene synonyms"
  },
  "length_gap_aa": 680,
  "length_defence_usable": true,
  "why_it_is_worse_than_an_ordinary_rule_62_case": "hERG/KCNH2 is the cardiac potassium channel behind QT prolongation, and QTc is a listed adverse event of the menin inhibitors this project is built around. 'ERG' in this corpus therefore denotes a transcription factor OR a drug-safety endpoint, in the same document set. Two entries of this graph already use hERG in the cardiac sense.",
  "token_hazard": "ERG is three letters occurring inside ordinary English words and surnames, and ERG is also the standard abbreviation for ELECTRORETINOGRAPHY and a unit of energy. ERG[tiab] returns 13,832 records of which the overwhelming majority are prostate cancer or ophthalmology. Never count ERG without title-checking."
}
domains
[
  {
    "name": "PNT (pointed / SAM domain)",
    "start": 113,
    "end": 199
  },
  {
    "name": "ETS DNA-binding domain",
    "start": 311,
    "end": 391
  }
]
shown for fusion
{
  "in_kmt2a_rearranged_context": "NO. No ERG experiment in a KMT2A-rearranged cell was found; the title-checked literature search returned none [E6] and DepMap is anti-selective [E5].",
  "in_another_fusion_context": "YES, several, all other diseases: FUS::ERG AML, EWS-ERG Ewing sarcoma, ETO2-GLIS2 (which upregulates and interacts with ERG at enhancers, per this graph's existing entries), and 3q26/EVI1-driven AML [E4].",
  "wild_type_only": "E1, E2, E3 are all wild-type systems - normal murine haematopoiesis and unselected adult AML cohorts.",
  "opposite_direction": "E7 - in DUX4-rearranged B-ALL the selected event is LOSS of ERG function."
}
expressed required maintains
{
  "expressed_in": "Blood stem/progenitor cells and AML blasts; high expression is a poor-prognosis risk factor in AML. Not measured in KMT2A-rearranged material.",
  "required_in": "Murine HSC self-renewal under stress only, not at steady state [E2]; EVI1-driven AML selectively [E4]. NOT required in KMT2A-rearranged lines - the effect runs the other way [E5].",
  "maintains_a_state": "The stressed self-renewing HSC state and, as a heptad member, the stem-cell programme that predicts AML outcome. In EVI1-driven AML, ERG maintains the undifferentiated state and its suppression causes terminal differentiation [E4]."
}
essentiality
{
  "source": "DepMap CRISPR (Chronos) via Open Targets GraphQL, retrieved live 2026-09-20",
  "n_all_lines": 1244,
  "chronos_mean_all_lines": -0.024,
  "frac_lines_at_or_below_minus_0_5": 0.0072,
  "mean_other_leukaemia": -0.1399,
  "mean_kmt2a_r_myeloid": 0.0214,
  "z_kmt2a_r_myeloid": 0.48,
  "mean_kmt2a_r_b_lineage": -0.2678,
  "z_kmt2a_r_b_lineage": -0.38,
  "positive_control_men1_z_myeloid": -2.41,
  "genome_scan_rank": 15074,
  "genome_scan_of": 17728,
  "genome_scan_difference": 0.0543,
  "interpretation": "ANTI-selective: ERG is less required in KMT2A-rearranged lines than in other leukaemia lines. Bounds a fitness dependency only; ERG's published leukaemia phenotype is differentiation, which this assay cannot see."
}
graph mention audit
{
  "date": "2026-09-20",
  "method": "word-boundary grep of data/graph/nodes for ERG",
  "node_files": 15,
  "registers_found": [
    "EWS-ERG Ewing sarcoma",
    "FUS::ERG AML",
    "DUX4-rearranged B-ALL / ERGalt / ERG deletion (loss of function)",
    "ETO2-GLIS2 upregulating and interacting with ERG at enhancers",
    "EVI1 acting via ERG",
    "fetal-versus-adult ontogeny permissiveness"
  ],
  "registers_about_a_kmt2a_axis": 0,
  "separate_herg_register": "2 node files use hERG in the cardiac QT-liability sense"
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.0838,
  "chronos_mean_other_leukaemia": -0.1381,
  "difference": 0.0543,
  "pan_essentiality_all_lines": -0.0242,
  "selectivity_rank": 15074,
  "rank_denominator": 17728,
  "provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
  "how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}
How this could bite us back
  • ERG is required for HSC self-renewal under transplantation and myelotoxic stress [E2] - exactly the condition a leukaemia patient's marrow is in during and after induction. An ERG-directed agent has its worst predicted toxicity in the setting where it would be used.
  • ERG is a tumour suppressor in at least one B-ALL subtype [E7]. An agent that lowers ERG function is, in that subtype, doing what the driver lesion already does.
  • The therapeutic name collision is a practical hazard, not a joke: 'ERG inhibition' and 'hERG inhibition' are one letter apart, and hERG inhibition is a cardiac safety liability on the drug class this project is centred on. Any document, protocol or label touching both needs the distinction written out.
  • There is no ERG-selective chemical matter. ETS-family transcription factors have resisted direct targeting for three decades, and the closest thing - the ETS-domain interface shared with FLI1 - is paralogue-blind in a family that includes the Ewing-sarcoma driver.
Still open
  1. Is ERG expressed at all in KMT2A-rearranged B-ALL, and is ERGalt ever present there? The DUX4 subtype shows ERG can be inactivated by an isoform switch invisible to standard expression analysis, and nobody has looked for ERGalt outside DUX4-rearranged cases.
  2. Does the EVI1-to-ERG axis operate in the EVI1-positive ~43% of MLL-rearranged AML? This is the same question gene-mecom asks and it is the one experiment that would put ERG properly into this graph as biology rather than as a comparator.
  3. Why does ERG-driven prostate cancer initiation require KMT2A and DOT1L (PMID 40858905)? A KMT2A/DOT1L requirement downstream of an ETS factor in a non-haematopoietic tissue is precisely the rule-77 case - a mechanism involving this graph's central protein published outside this graph's disease - and it has not been read.
  4. Does the heptad act as a unit in KMT2A-rearranged cells? ERG, GATA2 and LMO2 are all heptad members, all three are on this beat, and all three are non-selective or anti-selective in the KMT2A-rearranged DepMap arm. Either the heptad is not operating there, or it is operating and is buffered - and a combinatorial knockdown would distinguish those.
Every source behind this entry (7)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Genome-wide binding of ten key regulators of blood stem/progenitor cells (SCL/TAL1, LYL1, LMO2, GATA2, RUNX1, MEIS1, PU.1, ERG, FLI-1, GFI1B) revealed a previously unrecognised combinatorial interaction between a heptad of transcription factors - SCL, LYL1, LMO2, GATA2, RUNX1, ERG and FLI-1 - and implicated direct protein-protein interactions between RUNX1, GATA2, SCL and ERG in stabilising complex binding to DNA.

    Combinatorial transcriptional control in blood stem/progenitor cells: genome-wide analysis of ten major transcriptional regulators · Wilson NK, Foster SD, Wang X, Knezevic K, Schütte J, Kaimakis P, Chilarska PM, Kinston S, et al. · Cell Stem Cell 7:532-544 2010
    PMID 20887958DOI
    What this evidence cannot show

    Abstract fetched from PubMed efetch XML on 2026-09-20 and read in full; body not retrieved, so the cell system (a murine haematopoietic progenitor line in the original work) is NOT asserted beyond 'blood stem/progenitor cells' as the abstract phrases it, and the number of heptad-bound regions is not quoted. This is NORMAL haematopoiesis, not leukaemia, and there is no fusion anywhere in it. It is cited for one thing: that ERG, GATA2 and LMO2 are members of one combinatorially acting set, which is the reason three of this beat's six genes keep appearing in the same sentences in the literature.

    Evidence that argues against this (1)
    • A combinatorial binding set defined in normal stem/progenitor cells is a statement about normal regulatory architecture. Membership of it does not make a factor a leukaemia dependency, and in KMT2A-rearranged lines DepMap finds ERG, GATA2 and LMO2 all NON-selective or anti-selective, with GATA2 ranking 17,591 of 17,728 in the anti-selective direction. analysis run by this project 2026
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In mice, heterozygosity for functional Erg is sufficient for adequate steady-state HSC maintenance, but Erg(+/Mld2) mutants show impaired HSC self-renewal after bone-marrow transplantation or during recovery from myelotoxic stress, and Mpl(-/-) Erg(+/Mld2) compound mutants display exacerbated stem-cell deficiency and bone-marrow failure - so Erg is a critical regulator of adult HSCs essential for maintaining self-renewal at times of high HSC cycling.

    Erg is required for self-renewal of hematopoietic stem cells during stress hematopoiesis in mice · Ng AP, Loughran SJ, Metcalf D, Hyland CD, de Graaf CA, Hu Y, Smyth GK, Hilton DJ, et al. · Blood 118:2454-2461 2011
    PMID 21673349DOI
    What this evidence cannot show

    Abstract fetched from PubMed efetch XML on 2026-09-20 and read in full; body not retrieved. MURINE, NORMAL HAEMATOPOIESIS, and the allele is a point-mutant hypomorph (Mld2) in the heterozygous state rather than a clean null - so this bounds Erg's requirement at reduced dosage and does not establish what a complete loss does. Cited here for the normal-function baseline that any ERG-directed intervention would have to answer to, not as leukaemia evidence.

    Evidence that argues against this (1)
    • A hypomorphic heterozygote that is normal at steady state and fails only under stress says the system has substantial reserve. Read as a toxicity prediction, that cuts both ways: an ERG-directed agent might be tolerable at rest and catastrophic during marrow recovery from chemotherapy, which is exactly the setting a leukaemia drug is used in. Diffner E et al. 2013
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    ERG is a component of normal and leukaemic stem-cell signatures and high ERG expression is a risk factor for poor prognosis in AML; ERG expression in AML is associated with activity of the ERG promoters and a +85 stem-cell enhancer and with a heptad of transcription factors that combinatorially regulate genes in HSCs; and gene-expression signatures derived from that promoter/enhancer and heptad activity are associated with clinical outcome in cases where ERG expression alone fails.

    Activity of a heptad of transcription factors is associated with stem cell programs and clinical outcome in acute myeloid leukemia · Diffner E, Beck D, Gudgin E, Thoms JA, Knezevic K, Pridans C, Foster S, Goode D, et al. · Blood 121:2289-2300 2013
    PMID 23327922DOI
    What this evidence cannot show

    Abstract fetched from PubMed efetch XML on 2026-09-20 and read in full; body not retrieved, so the cohorts, the signature genes and the hazard ratios are NOT asserted. The abstract does not enumerate the heptad members - that enumeration is taken from E1, which defined the set. ADULT AML COHORTS, no KMT2A-rearranged stratum named in the abstract, no paediatric or infant material, no B-lineage. A prognostic association is not a dependency and this paper makes no functional claim.

    Evidence that argues against this (1)
    • The same abstract reports the heptad signature is associated with AMLs LACKING NPM1 mutations and confers adverse prognosis when combined with FLT3 mutations - so the prognostic signal is conditioned on co-mutation status, and a prognostic marker that behaves differently in different genotypes is a poor guide to biology in a genotype it was not measured in. KMT2A-rearranged AML is one such genotype. Wilson NK et al. 2010
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In human and murine EVI1-driven AML, ERG is a direct transcriptional target of EVI1 that is aberrantly expressed and selectively required; EVI1 occupies a conserved intragenic ERG enhancer in AML cell lines and primary patient samples; suppression of ERG induces terminal differentiation of EVI1-driven AML cells; and ectopic expression of ERG abrogates their dependence on EVI1.

    EVI1 drives leukemogenesis through aberrant ERG activation · Schmoellerl J, Barbosa IAM, Minnich M, Andersch F, Smeenk L, Havermans M, Eder T, Neumann T, et al. · Blood 141:453-466 2023
    PMID 36095844DOI
    What this evidence cannot show

    Abstract fetched from PubMed efetch XML on 2026-09-20 and read in full; body not retrieved. THE SYSTEM IS 3q26/EVI1-DRIVEN AML, NOT KMT2A-REARRANGED AML - rule 73. It is included here because it is the single strongest functional statement about ERG in any acute leukaemia and because MECOM is on this same beat, not because it transfers. The transfer to the EVI1-positive subset of MLL-rearranged AML is untested.

    Evidence that argues against this (1)
  5. E5established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    Fetched live on 2026-09-20, DepMap Chronos gene effect for ERG has mean -0.024 across 1,244 cell lines with 0.72% of lines at or below -0.5; among leukaemia lines the KMT2A-rearranged myeloid half sits at z +0.48 and the B-lineage half at z -0.38 relative to 44 other leukaemia lines whose own mean is -0.14, and this project's genome-wide scan ranks ERG 15,074 of 17,728 with a difference of +0.0543 - i.e. in the ANTI-selective direction. MEN1 through the identical query shape returned z -2.41 / -2.43.

    What this evidence cannot show

    A bounded negative with a date, a route and a positive control through the identical query shape (rule 67). Seven myeloid and four B-lineage KMT2A-rearranged lines; the B-lineage z of -0.38 rests on four lines of which HB11;19 at -0.485 is the only mover, so it is one line rather than a result. None of the 44 comparator leukaemia lines is annotated for EVI1 status, which is the stratification [E4] says matters. Immortalised adult-derived lines, none infant, 21-day fitness read-out with no differentiation or self-renewal endpoint - and [E4]'s ERG phenotype is DIFFERENTIATION, which this assay cannot see. Ensembl id from HGNC primary (HGNC:3446); approvedSymbol read back as ERG so the KCNH2 collision could not enter.

    Evidence that argues against this (1)
    • A fitness screen cannot see a differentiation phenotype, and ERG's strongest published leukaemia phenotype is exactly that. The anti-selective ranking is a real measurement and it bounds ERG as a KMT2A-selective FITNESS dependency only; it does not bound ERG as a differentiation-state factor in this disease, which nobody has measured. Schmoellerl J et al. 2023
  6. E6moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    Counted absence, measured 2026-09-20 with PubMed E-utilities esearch: ERG[tiab] returns 13,832 records and (MLL OR KMT2A)[tiab] returns 9,228, while ERG[tiab] AND (MLL OR KMT2A)[tiab] AND leukemia returns 63. Title-checking the 20 most recent of those 63 found fusion-detection and RNA-sequencing panels, FUS::ERG AML, KMT2A-rearranged sarcoma, an EVI1-via-ERG chemoresistance paper, and one prostate-cancer paper - and no paper about ERG as a cofactor of a KMT2A fusion.

    ERG-driven prostate cancer initiation is cell-context dependent and requires KMT2A and DOT1L · Feng W, Ladewig E, Lange M, Salsabeel N, Zhao H, Lee YS, Gopalan A, Luo H, et al. · Nature Genetics 57:2177-2191 2025
    PMID 40858905DOI
    What this evidence cannot show

    GRADED MODERATE BECAUSE ONLY 20 OF 63 RECORDS WERE TITLE-CHECKED - a partial title check, reported as such. Bounded to titles and abstracts; supporting figures not searched (rule 80). The ERG operand of 13,832 is dominated by prostate cancer and by electroretinography, for which ERG is also the standard abbreviation, so the raw operand overstates the relevant literature by a large and unmeasured factor. NOTE ON THE CITATION SHAPE: the citation record is the Nature Genetics paper the search surfaced, with metadata copied field-by-field from efetch, and the query text sits in citation.query_note; it was first written with the query text in the title and authors fields beside that PMID, which conflates a citation with a query result and which verify-citations.py --strict correctly flagged. RULE 77 NOTE: that paper is in PROSTATE CANCER - ERG-driven initiation there requires KMT2A and DOT1L - which is precisely the case rule 77 warns about, a KMT2A mechanism published outside this graph's disease. Its TITLE was fetched and read; its abstract was NOT, and no claim is made from it here beyond the title.

    Evidence that argues against this (1)
    • A partial title check of a 63-record set is a weak instrument and the 43 unchecked records could contain the paper this search was looking for. The stronger evidence against an ERG-KMT2A cofactor relationship is the DepMap anti-selectivity, which is a measurement rather than an absence. analysis run by this project 2026
  7. E7established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    In a B-progenitor ALL subtype comprising up to 7% of B-ALL, DUX4 rearrangement and overexpression is present in all cases and is accompanied by transcriptional deregulation of ERG, expression of a novel ERG isoform ERGalt whose non-canonical first exon is transcribed from a DUX4 binding site, and frequent ERG deletion; ERGalt retains ERG's DNA-binding and transactivation domains but inhibits wild-type ERG transcriptional activity and is transforming - a paradigm in which a transcription-factor deregulation results in LOSS of ERG function.

    Deregulation of DUX4 and ERG in acute lymphoblastic leukemia · Zhang J, McCastlain K, Yoshihara H, Xu B, Chang Y, Churchman ML, Wu G, Li Y, et al. · Nature Genetics 48:1481-1489 2016
    PMID 27776115DOI
    What this evidence cannot show

    Abstract fetched from PubMed efetch XML on 2026-09-20 and read in full; body not retrieved. HUMAN, B-LINEAGE, PAEDIATRIC/AYA ALL - the right lineage and the right age, and the WRONG driver: this is DUX4-rearranged ALL, not KMT2A-rearranged. It is included because it is the only well-powered statement about ERG in B-ALL and because its DIRECTION is the opposite of the cooperating-factor premise that sent this beat looking. It does not test ERG in KMT2A-rearranged B-ALL and nothing here should be read as doing so.

    Evidence that argues against this (1)
    • ERG loss being required in one B-ALL subtype says nothing directly about another. ERG is a context-dependent factor - required for HSC self-renewal under stress, oncogenic when overexpressed in AML, and a tumour suppressor whose loss is selected for in DUX4-rearranged B-ALL - so no single sign can be carried across subtypes, including into KMT2A-rearranged disease where it has not been measured. Ng AP et al. 2011
Connected to