Oncofusion

start / graph / gene-setd1a

Proteinstrong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.Q5c-interactome2

SETD1A — a leukemia dependency whose enzyme activity turns out to be dispensable

also called SETD1A · SET1A · SET domain containing 1A · KMT2F

SETD1A is a separate H3K4-writing enzyme that shares the same four-protein core as KMT2A. Leukemia cells need it — but not for its enzyme activity, which can be deleted without hurting them, so the usual assumption about why it matters is wrong.

The same thing, for a scientist

SETD1A (O15047, 1,707 aa) is a SET1-family H3K4 methyltransferase, not a component of the KMT2A-fusion complex; it shares only the WRAD core. It is an AML dependency, and Hoshii and colleagues showed that dependency is catalytically independent: the SET domain is dispensable and a distinct FLOS domain recruits cyclin K for the DNA damage response. SETD1A is DepMap common-essential (-0.911, 94.7% of lines) and heavily constrained (LOEUF 0.264). It shows a nominal leukemia-restricted correlation with MEN1 of r = 0.359 that fails Holm correction.

What it looks like
SETD1A1,707 aa
RRMRRM · residues 84–172HCFC1-binding motif (HBM) · residues 1299–1303WDR5-interaction (WIN) motif · residues 1492–1497RxxxRR motif · residues 1537–1542SET SET domain (dispensable for AML survival) · residues 1568–1685Post-SET · residues 1691–170705001,0001,500
loading AF-O15047-F1…
AlphaFold AF-O15047-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
The result that overturns the obvious interpretation

The natural story about SETD1A in leukemia is that it writes H3K4me3 and therefore sustains the fusion's target genes. That story is contradicted by the paper that defined the dependency.

Hoshii and colleagues showed that SETD1A's methyltransferase activity is not required for AML cell survival. The SET domain can be deleted. What is required is a distinct region, the FLOS domain, which recruits cyclin K and supports the DNA damage response 1.

Any node in this graph asserting that SETD1A H3K4 methylation sustains fusion target genes is contradicted by this result and should be corrected. The practical consequence is larger than a wording change: a catalytic inhibitor of SETD1A would not reproduce the dependency, so the entire SET-domain-inhibitor approach to this protein is aimed at the wrong function.

The relationship to KMT2A is through WRAD, not through the fusion

SETD1A shares the WRAD core with MLL1 and the other KMT2 enzymes, and uses the same conserved WIN motif to engage WDR5 — UniProt places the SETD1A WIN motif at residues 1492–1497, and the WDR5–SETD1A WIN peptide complex is solved at 1.50 Å (4EWR) 34.

That shared architecture has a consequence worth recording: WIN-motif inhibitors developed against MLL1 also inhibit SETD1A, because complexes with the weakest Win-motif-WDR5 interaction are the most susceptible, and MLL1 and SETD1A are the two that qualify 2. A WDR5 WIN-site drug is therefore not an MLL1-selective agent; it is an MLL1-plus-SETD1A agent. Since the KMT2A fusion has lost the SET domain and no longer needs WDR5 in that way, a WIN-site inhibitor in KMT2A-rearranged leukemia would be acting on wild-type COMPASS and SETD1A rather than on the fusion.

Dependency and tolerability

SETD1A is DepMap common-essential: mean Chronos -0.911, leukemia -0.874, depleted in 94.7% of 1,258 lines 5. gnomAD: pLI 1.0, LOEUF 0.264, 29 observed against 149.9 expected loss-of-function variants — strong, well-powered constraint 5.

Leukemia-restricted correlation with MEN1 is r = 0.359 (95% CI 0.115–0.586, permutation p = 0.007), which fails Holm correction across 16 genes 6. Note that this is a pan-essential gene producing a non-zero correlation, which is worth flagging: it weakens this project's general claim that flat profiles cannot correlate. The correlation should not be acted on, but the methodological point stands.

Expression is flat and low, with bone marrow at 20.2 nTPM and both classical monocytes (0.2) and naive B cells (0.5) close to zero 5. No verified mouse knockout; we searched specifically for a Setd1a haematopoietic knockout and did not confirm one, so none is asserted here.

Facts and figures
uniprotO15047
gene symbolSETD1A
ensemblENSG00000099381
length AA1707
alphafold IDAF-O15047-F1
complexSET1A/COMPASS; shares the WRAD core with MLL1 but is not a component of the KMT2A-fusion complex
key correctionThe AML dependency on SETD1A is catalytically independent: the SET domain is dispensable and the required function is FLOS-domain recruitment of cyclin K for the DNA damage response.
druggabilityA SET-domain catalytic inhibitor would not reproduce the dependency. The FLOS-cyclin K interface is the functionally relevant surface and has no structure and no chemical matter. WIN-site inhibitors hit SETD1A alongside MLL1 and are therefore not fusion-selective.
cell context evidence classrequired_in (AML cells, domain-resolved) + maintains_a_state (mitochondrial respiration and S-phase DDR in AML; quiescence in neural stem cells). Kept separate: the AML requirement is a survival readout, the respiration result is a metabolic-state readout, and the neural result is a different tissue entirely.
cell context shown for fusionNOT ESTABLISHED FROM THE ABSTRACTS, and stated that way rather than guessed. Both load-bearing papers say 'AML cells' without naming the lines or their fusion status in the abstract, so the graph cannot record a KMT2A-rearranged context for them. Measured 2026-09-21: SETD1A[tiab] AND (KMT2A[tiab] OR MLL[tiab] OR "MLL-AF9"[tiab] OR MOLM[tiab]) returns 22 records, so a fusion-context literature does exist and several titles are explicitly MLL-rearranged - but the specific systems used in PMID 29474905 and PMID 36450243 were not resolved here. Zero in titles and abstracts is NOT the claim; the claim is that the system is unresolved and the supporting figures and methods were not read.
cell context depmapMildly toward the index arm but close to pan-essential. SETD1A is -0.949 in the 11 KMT2A-rearranged lines against -0.849 in 43 other leukaemia lines, difference -0.100, rank 1,146 of 17,728 (6.46th percentile), on a pan-line Chronos mean of -0.911. Substantially required everywhere, so this is a small differential on a large general requirement and the floor-compression caution applies.
cell context druggability consequenceThe SET domain is dispensable for AML cell survival and the FLOS/cyclin-K site is not. Any SETD1A strategy aimed at the methyltransferase activity is therefore aimed at the wrong part of the protein for this phenotype, which is unusual and worth carrying forward: a catalytic inhibitor would be expected to fail where a degrader or a cyclin-K interface disruptor would work.
subnuclear localizationchromatin at transcriptional start sites, tethered by BOD1L through the SETD1A FLOS domain; loss of BOD1L strips SETD1A from TSS without touching H3K4me3 (PMID 38989615). UniProt additionally annotates a nuclear-speckle pool (O15047, ECO:0000269) with no leukaemia source behind it. No image of SETD1A in any haematopoietic cell has been published.
shown for fusionNO. SETD1A has never been localised in a KMT2A-rearranged cell. Measured 2026-09-21 with NCBI esearch: (SETD1A[tiab] OR SET1A[tiab] OR KMT2F[tiab]) AND (MLL[tiab] OR KMT2A[tiab] OR MLL-AF4[tiab] OR MLL-AF9[tiab] OR mixed lineage leukemia[tiab]) AND (localiz*[tiab] OR localis*[tiab] OR subcellular[tiab] OR nuclear body[tiab] OR nuclear bodies[tiab] OR speckle*[tiab] OR condensat*[tiab] OR nucleolar[tiab] OR immunofluorescence[tiab] OR compartment*[tiab]) returns 2 records, both title-checked off-topic (PMID 38266981, a KMT2-family review; PMID 24680668, a SET-domain surface structural study). Operands: gene clause 300, gene AND fusion 45, so the intersection is real and the zero falls at the localisation step. Positive control through the identical query shape, with (MLLT1[tiab] OR ENL[tiab]) substituted for the gene clause, returns 23 - the instrument can return a one. Zero in titles and abstracts; supporting figures not searched (brief rule 80).
domains
RRM84.0RRM: 84HCFC1-binding motif (HBM)1,299HCFC1-binding motif (HBM): 1299WDR5-interaction (WIN) mo…1,492WDR5-interaction (WIN) motif: 1492RxxxRR motif1,537RxxxRR motif: 1537SET domain (dispensable f…1,568SET domain (dispensable for AML survival): 1568Post-SET1,691Post-SET: 1691
PDB IDS
[
  "3S8S",
  "3UVN",
  "4EWR",
  "8ILY"
]
PDB verified
[
  "4EWR",
  "8ILY"
]
domains — the numbers behind the chart
[
  {
    "name": "RRM",
    "start": 84,
    "end": 172,
    "pfam": null
  },
  {
    "name": "HCFC1-binding motif (HBM)",
    "start": 1299,
    "end": 1303,
    "pfam": null
  },
  {
    "name": "WDR5-interaction (WIN) motif",
    "start": 1492,
    "end": 1497,
    "pfam": null
  },
  {
    "name": "RxxxRR motif",
    "start": 1537,
    "end": 1542,
    "pfam": null
  },
  {
    "name": "SET domain (dispensable for AML survival)",
    "start": 1568,
    "end": 1685,
    "pfam": null
  },
  {
    "name": "Post-SET",
    "start": 1691,
    "end": 1707,
    "pfam": null
  }
]
essentiality
{
  "depmap_common_essential": true,
  "chronos_mean": -0.9112,
  "chronos_mean_leukemia_lines": -0.8743,
  "n_lines": 1258,
  "fraction_lines_depleted": 0.947,
  "gnomad_pLI": 1,
  "gnomad_loeuf": 0.2638,
  "gnomad_obs_lof": 29,
  "gnomad_exp_lof": 149.9,
  "mouse_ko": "Not established; no Setd1a knockout citation verified and none asserted",
  "hpa_ntpm": {
    "bone_marrow": 20.2,
    "testis": 13.4,
    "thymus": 9.6,
    "lymph_node": 9.3,
    "spleen": 8.6,
    "heart_muscle": 5.4,
    "liver": 4.3,
    "naive_B_cell": 0.5,
    "classical_monocyte": 0.2
  },
  "hpa_specificity": "Low tissue specificity"
}
codependency men1
{
  "r_leukemia_lines": 0.3593,
  "ci_leukemia": [
    0.1154,
    0.5864
  ],
  "crosses_zero": false,
  "p_permutation": 0.0068,
  "p_holm": 0.0884,
  "survives_holm": false
}
cell type
{
  "required_in": "AML cells - and required through a NON-CATALYTIC function, which is the part that matters for how this gene would be drugged. Assessment of the MLL/SET family found SETD1A required for AML cell survival; mutagenesis and CRISPR-Cas9 domain screening showed the enzymatic SET domain is NOT necessary for AML cell survival, while a distinct region termed the FLOS (functional location on SETD1A) domain is indispensable. FLOS disruption suppresses DNA-damage-response genes and induces p53-dependent apoptosis, and the FLOS domain is a cyclin-K-binding site required for chromosomal recruitment of cyclin K and for DNA-repair-associated gene expression in S phase.",
  "how_shown": "CRISPR-Cas9 domain screening, mutagenesis, targeted protein degradation, CRISPR functional analysis and metabolomics, in AML cell lines",
  "systems": "AML cell lines. The abstracts do not name the lines or their fusion status, so this cannot be attributed to a KMT2A-rearranged context from the abstract alone.",
  "nodes": []
}
cell state
{
  "maintains_state": "Two states, both non-catalytic. (1) S-phase DNA-damage-response competence, through cyclin K recruitment. (2) MITOCHONDRIAL RESPIRATION: rapid degradation of SETD1A by a degrader immediately downregulates transcripts of DNA-repair and heme-biosynthesis pathways, and CRISPR functional analysis with metabolomics reveals an essential role for SETD1A in maintaining mitochondrial respiration in AML cells, with the targets enriched in head-to-head gene pairs and the mechanism running through strand-specific promoter-proximal RNAPII pausing.",
  "cross_lineage_state": "In mouse, Setd1a maintains adult neural stem cell QUIESCENCE in the hippocampus - a maintains-a-state result in a completely different tissue, recorded because it says the gene holds cell states as a general matter and because it is an off-target surface for any SETD1A-directed agent.",
  "nodes": [
    "cell-state-oxphos-dependence",
    "cell-state-quiescence"
  ]
}
localization
{
  "subcellular": "Nucleus, with an experimentally annotated cytoplasmic pool. UniProt O15047 lists Nucleus speckle, Chromosome and Cytoplasm, all at ECO:0000269. Nothing in the leukaemia literature images SETD1A at all; the leukaemia-relevant measurements are genomic occupancy, not microscopy.",
  "subnuclear": "CHROMATIN, at transcriptional start sites, and held there by a separate protein. BOD1L knockout immediately reduces SETD1A at TSS, causes an elongation defect and raises Pol II at TSS without reducing H3K4me3; BOD1L's Shg1 domain binds DNA and its W104 residue recruits SETD1A through the SETD1A FLOS domain. That same FLOS domain is the cyclin-K-binding site required for chromosomal recruitment of cyclin K, so SETD1A's subnuclear address and its non-catalytic cargo run through one module.",
  "how_shown": "ChIP occupancy at TSS plus domain mapping and DNA-binding assays in BOD1L-knockout human cancer and leukaemia cells (PMID 38989615); CRISPR-Cas9 domain screening and chromosomal recruitment assays for the FLOS-cyclin K axis (PMID 29474905); UniProt O15047 SUBCELLULAR LOCATION comment for the compartment list.",
  "systems": "human cancer and leukaemia cell lines of unstated fusion status; no primary patient material and no KMT2A-rearranged arm in either localisation source",
  "node": "localization-chromatin-bound-versus-soluble-fraction-kmt2a"
}
How this could bite us back
  • Essential-function toxicity: required by 94.7% of cell lines and strongly LoF-constrained in humans.
  • Wrong-function risk: the field's instinct is to inhibit the methyltransferase, and the methyltransferase activity is dispensable. Chemistry aimed at the SET domain would generate clean pharmacology against an irrelevant function.
  • Selectivity leak: WIN-motif inhibitors intended for MLL1 also inhibit SETD1A, so a WDR5-directed agent carries SETD1A pharmacology whether or not that was intended.
  • ESSENTIAL-FUNCTION TOXICITY — SETD1A is close to pan-essential in DepMap (-0.911 across 1,208 lines) and maintains quiescence in adult neural stem cells in mouse. A degrader - which is what the non-catalytic mechanism implies you would need, since the SET domain is dispensable - removes the whole protein and therefore all of these functions at once, which is a wider perturbation than a catalytic inhibitor would deliver.
  • ESSENTIAL-FUNCTION TOXICITY — The AML survival phenotype is p53-dependent apoptosis downstream of DNA-damage-response gene suppression. In TP53-mutant or TP53-deleted disease the phenotype should be blunted, and TP53 loss is a known feature of relapsed and therapy-related leukaemia - so the patients most in need of a new agent are the ones in whom this mechanism is predicted to work least well.
Still open
  1. Is the SETD1A dependency catalytically independent in KMT2A::AFF1 B-ALL as it is in AML? All of the evidence is myeloid, and a lymphoid confirmation would be cheap.
  2. Does the FLOS-cyclin K interface have a structure? Without one there is no route to the function that actually matters, and the four existing SETD1A PDB entries all cover other regions.
  3. Would a WDR5 WIN-site inhibitor in KMT2A-rearranged leukemia be acting on the fusion at all? The fusion has lost the SET domain, so the WIN interaction it would disrupt belongs to wild-type MLL1 and SETD1A. This deserves an explicit answer before any WIN-site compound is proposed for this disease.
  4. Which AML lines were used in the two load-bearing SETD1A studies, and were any of them KMT2A-rearranged? Both abstracts say only 'AML cells'. This is resolvable from the methods sections in an afternoon and it determines whether this entry's central claims are in-context or borrowed.
  5. SETD1A maintains mitochondrial respiration in AML cells, and this graph records OXPHOS dependence as inverted between KMT2A-rearranged myeloid and lymphoid disease. Does SETD1A degradation lower or raise respiration in a KMT2A-rearranged B-ALL line? If it raises leukaemia-initiating capacity in the lymphoid setting, the therapeutic proposal in this entry is lineage-specific in a way nobody has stated.
  6. Setd1a maintains adult neural stem cell quiescence in mouse. Does SETD1A loss wake, or deepen, the quiescent leukaemic persister state that this graph invokes to explain measurable residual disease?
  7. Is SETD1A's TSS occupancy in a KMT2A::AFF1 cell BOD1L-dependent in the same way it is in the leukaemia lines of PMID 38989615, and does BOD1L knockout displace SETD1A from the HOXA cluster specifically? One ChIP in an index line would convert a general chromatin address into a fusion-relevant one.
  8. Nobody has photographed SETD1A in a blood cell. Does the UniProt nuclear-speckle annotation reproduce in a KMT2A-rearranged line, or is it a property of the cell type in which it was first seen?
Every source behind this entry (11)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    SETD1A is required for AML cell survival through a non-catalytic function: the SET domain is dispensable, and a distinct FLOS domain recruits cyclin K and supports the DNA damage response.

    A Non-catalytic Function of SETD1A Regulates Cyclin K and the DNA Damage Response. · Hoshii T, Cifani P, Feng Z, Huang CH, Koche R, Chen CW, Delaney CD, Lowe SW, Kentsis A, Armstrong SA · Cell 2018
    PMID 29474905DOI
    What this evidence cannot show

    Mouse and human AML models; the work is myeloid and does not address KMT2A::AFF1 infant B-ALL. A dispensable SET domain in a CRISPR-rescue experiment does not exclude a contribution of catalysis that becomes visible under other conditions or over longer timescales.

    Evidence that argues against this (3)
    • The same non-catalytic SETD1A function is required by normal haematopoietic stem cells, which removes most of the therapeutic window. Conditional Setd1a deletion in adult long-term HSCs abolishes the expression of DNA damage recognition and repair genes, and although phenotypic LT-HSC numbers in the marrow are largely preserved, the cells lose their transcriptional identity and lose proliferative capacity and stem-cell function under replicative stress in situ and after transplantation; SETD1A also protects HSPCs from activation-induced attrition in response to inflammatory stimulation. A paediatric leukemia patient is under exactly these stresses. This is the archetypal essential-function toxicity: the DNA-damage-response programme SETD1A supports is the same one in the malignant clone and in the stem cell that has to rebuild the marrow. Arndt K et al. 2018
    • The FLOS function is not autonomous to SETD1A. BOD1L is the highest-correlated SETD1A co-dependency across human cancer cell lines, and a single tryptophan (W104) in the BOD1L Shg1 domain recruits SETD1A to chromatin by binding the FLOS domain. BOD1L knockout reduces leukemia cell number in vitro and in vivo and phenocopies the SETD1A-knockout transcriptional profile, while SETD1A occupancy at transcription start sites collapses without any reduction in H3K4me3. So the interface that matters is SETD1A-BOD1L, not SETD1A alone, and an agent aimed at SETD1A may be aiming at the wrong half of a two-protein module. It also means E1's framing of the FLOS domain as an intrinsic SETD1A activity is incomplete. Hoshii T et al. 2024
    • When SETD1A is removed by acute degradation rather than by mutagenesis, the immediate transcriptional collapse is in DNA repair AND heme biosynthesis, and the requirement traces to mitochondrial respiration. That reframes the dependency as a general metabolic and housekeeping one in AML rather than a KMT2A-fusion-specific one, and it also shows that the phenotype is only obtained by removing the protein: an occupancy-based inhibitor of a domain that has no catalytic activity has nothing to inhibit. The tractability implied by E1 is therefore degrader tractability, not inhibitor tractability. Hoshii T et al. 2022
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Disrupting the Win-motif-WDR5 interaction specifically inhibits MLL1 and SETD1A core complexes while only modestly disrupting MLL2/MLL4 and SETD1B, because complexes with the weakest Win-motif-WDR5 interaction are most susceptible to Win-motif-based inhibitors.

    PMID 27563068DOI
    What this evidence cannot show

    Reconstituted complexes and a peptidomimetic in vitro, not cells or animals. The inference that a WIN-site drug would act on wild-type COMPASS rather than on the fusion is ours and follows from the fusion having lost the SET domain.

    Evidence that argues against this (2)
    • The premise that WDR5-Win-motif inhibition is worth doing at all in KMT2A-fusion disease is undercut by the genetics of MLL1 catalysis. The SET domain of MLL1 is dispensable both for maintaining haematopoietic stem cells and for supporting MLL-AF9-driven leukemogenesis in mouse; what matters at MLL1 target genes is recruited MOF and H4K16 acetylation, not MLL1's own methyltransferase output. A Win-motif inhibitor works by disassembling the module that stimulates SET-domain catalysis. If that catalysis is not required for leukemogenesis, the mechanism of action asserted for these compounds is not the mechanism by which they would need to work, and any activity they show is likely to be off the stated target. Mishra BP et al. 2014
    • Within the same study that supports E2, a paralogue escape route is reported: in the absence of WDR5, MLL3 still interacts with RAD and shows ENHANCED activity. A Win-motif peptidomimetic that selectively disables MLL1 and SETD1A therefore selects for a chromatin state in which a WDR5-independent SET1-family enzyme is more, not less, active. That is a compensation mechanism built into the target class, and it belongs in the bite-back reasoning for any WDR5-interface strategy. Alicea-Velazquez NL et al. 2016
  3. E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProt O15047 annotates human SETD1A as 1,707 residues with an RRM at 84-172, an HCFC1-binding motif at 1299-1303, a WDR5-interaction (WIN) motif at 1492-1497, an RxxxRR motif at 1537-1542, a SET domain at 1568-1685 and a post-SET region at 1691-1707.

    What this evidence cannot show

    The FLOS domain identified by Hoshii and colleagues as the functionally required region is not annotated as a feature in this UniProt record, so its coordinates should be read from the paper rather than from these boundaries.

    Evidence that argues against this (1)
    • A UniProt motif table is curated annotation, and the WIN motif it lists at 1492-1497 is the point of interest: SETD1A retains a WDR5-binding motif that KMT2A fusions lose, so the two proteins compete for WDR5 in a way the fusion cannot. UniProt Consortium 2026
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    Verified structures include the isolated SETD1A RRM domain at 1.70 A (8ILY) and the WDR5-SETd1a WIN motif peptide binary complex at 1.50 A (4EWR).

    What this evidence cannot show

    Only four PDB entries cover SETD1A and none covers the FLOS region that carries the required function. The structural biology and the functional biology of this protein are looking at different parts of it.

    Evidence that argues against this (1)
    • Counting PDB entries measures how tractable a protein was to CRYSTALLISE, which correlates with being small, ordered and well-behaved rather than with being druggable. What decides design value is whether the deposited density covers the site that matters, at a resolution that supports modelling, in the relevant state — and this graph has repeatedly found structures described in the direction that overstates what is resolved. An isolated RRM and a WIN-motif peptide binary complex are two small fragments of a 1,707-residue protein. RCSB PDB 2026
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    SETD1A is DepMap common-essential with mean Chronos -0.911 across 1,258 lines and depletion in 94.7% of lines; gnomAD reports pLI 1.0 and LOEUF 0.264 with 29 observed against 149.9 expected loss-of-function variants; HPA reports bone marrow 20.2 nTPM with classical monocytes at 0.2 and naive B cells at 0.5.

    What this evidence cannot show

    DepMap release version not stated by the API.

    Evidence that argues against this (1)
    • Pan-essentiality is the finding here and it argues against the target rather than for it. A gene depleted in more than 90% of all cell lines is required by the patient's normal marrow as much as by the leukaemia, so its appearance in a KMT2A complex is a statement about the complex and not a therapeutic opening. gnomAD, Broad Institute 2026
  6. E6weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.dataset

    Across 54 DepMap leukemia lines SETD1A correlates with MEN1 at r = 0.359 (95% CI 0.115-0.586, permutation p = 0.007), failing Holm correction across 16 genes (Holm p = 0.088), which is notable because a pan-essential gene is not expected to produce a non-zero correlation at all.

    Leukemia-restricted co-dependency with MEN1, bootstrap and permutation testing; analyses/wrad-discrimination/ in this project · Broad Institute DepMap; Open Targets Platform; analysis run by this project · Database and derived analysis 2026
    What this evidence cannot show

    Our own computation in 54 lines, failing correction. Do not cite SETD1A as a co-dependency. Even pan-essential genes retain some residual variance, so a nominal correlation is not evidence against the flat-profile argument so much as a reminder that it is a tendency rather than a rule.

    Evidence that argues against this (1)
    • Same objection as for the other sub-threshold correlations in this beat: the deeper 2,743-gene variance-matched null with Holm correction leaves only KMT2A, PSIP1, DOT1L, FLT3, MEF2C and KAT7 standing, with MEIS1 already failing at Holm 0.067. LEO1's r = 0.302 at Holm 0.286 in the 16-gene panel is well below the genes that fail in the deep pool, so it should not be described as a signal of any kind. Broad Institute DepMap; Open Targets Platform; analysis run by this project 2026
  7. E7strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    SETD1A is required for the survival of acute myeloid leukaemia cells through a non-catalytic function: the enzymatic SET domain is not necessary for AML cell survival, while a distinct FLOS (functional location on SETD1A) domain is indispensable; FLOS disruption suppresses DNA-damage-response genes and induces p53-dependent apoptosis, and the FLOS domain acts as a cyclin-K-binding site required for chromosomal recruitment of cyclin K and for DNA-repair-associated gene expression in S phase.

    A Non-catalytic Function of SETD1A Regulates Cyclin K and the DNA Damage Response. · Hoshii T, Cifani P, Feng Z, Huang CH, Koche R, Chen CW, Delaney CD, Lowe SW, Kentsis A, Armstrong SA · Cell 2018
    PMID 29474905DOI
    What this evidence cannot show

    AML cell lines with CRISPR-Cas9 domain screening and mutagenesis; the abstract does not name the lines or their fusion status, so no KMT2A-rearranged attribution is made here. No B-lineage, no infant and no primary-patient arm in the abstract. The p53-dependence of the apoptosis means the phenotype is genotype-conditional and would be expected to differ in TP53-mutant disease. Read: abstract only.

    Evidence that argues against this (1)
  8. E8strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    SETD1A maintains the metabolic state of AML cells: rapid degradation of SETD1A by a protein-degrader technology immediately downregulates transcripts associated with DNA repair and heme biosynthesis, and CRISPR-based functional analysis with metabolomics reveals an essential role for SETD1A in maintaining mitochondrial respiration in AML cells, with targets enriched in head-to-head gene pairs and the mechanism acting through a non-enzymatic, cyclin-K-dependent effect on strand-specific promoter-proximal RNA polymerase II pausing.

    SETD1A regulates transcriptional pause release of heme biosynthesis genes in leukemia. · Hoshii T, Perlee S, Kikuchi S, Rahmutulla B, Fukuyo M, Masuda T, Ohtsuki S, Soga T, Nabet B, Kaneda A · Cell Rep 2022
    PMID 36450243DOI
    What this evidence cannot show

    AML cell lines with a degron system; the abstract does not name the lines or their fusion status. Mitochondrial respiration is inferred from CRISPR functional analysis plus metabolomics rather than from direct respirometry in the abstract. Read: abstract only. This graph's OXPHOS entry records the dependence as established in KMT2A-rearranged AML and INVERTED in KMT2A-rearranged B-ALL, so the lineage in which this SETD1A result was obtained matters a great deal and is not stated.

    Evidence that argues against this (1)
    • The direction of the OXPHOS link cannot be carried across lineages in this disease. This graph's own OXPHOS cell-state entry records that KMT2A-rearranged myeloid leukaemia depends on mitochondrial respiration while in the lymphoid form OXPHOS inhibition instead makes leukaemia-initiating cells appear. A SETD1A-maintains-respiration result from unnamed AML lines therefore predicts nothing for infant B-ALL and might predict the opposite. Hoshii T et al. 2018
  9. E9moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    SETD1A holds cell states outside the haematopoietic system as well: in the mouse hippocampus, Setd1a is required for the epigenetic maintenance of adult neural stem cell quiescence.

    Epigenetic maintenance of adult neural stem cell quiescence in the mouse hippocampus via Setd1a. · Zhao T, Hong Y, Yan B, Huang S, Ming GL, Song H · Nat Commun 2024
    PMID 38971831DOI
    What this evidence cannot show

    Mouse hippocampal neural stem cells - a completely different tissue from anything else in this entry, recorded because it establishes that SETD1A maintains cell states generally and because it is an off-target surface for a SETD1A degrader. Read: abstract only. Nothing here is haematopoietic or leukaemic.

    Evidence that argues against this (1)
    • The neural result is about maintaining quiescence while the AML result is about maintaining proliferative survival and mitochondrial respiration, so SETD1A's state-maintaining function points in opposite directions in the two tissues. That is a reason to be careful about predicting the consequence of SETD1A loss in any untested cell type, including the quiescent leukaemic persister this graph tracks separately. Hoshii T et al. 2022
  10. E_q66_loc1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    In human leukaemia and cancer cell lines, SETD1A's chromatin address is delegated to BOD1L rather than intrinsic: BOD1L knockout immediately reduced SETD1A distribution at transcriptional start sites, induced a transcriptional elongation defect and increased RNA polymerase II content at TSS - without reducing H3K4me3 - and the Shg1 domain of BOD1L has DNA-binding ability, with a tryptophan residue (W104) recruiting SETD1A to chromatin through the SETD1A FLOS domain.

    BOD1L mediates chromatin binding and non-canonical function of H3K4 methyltransferase SETD1A. · Hoshii T; Kikuchi S; Kujirai T; Masuda T; Ito T; Yasuda S; Matsumoto M; Rahmutulla B; Fukuyo M; Murata T; Kurumizaka H; Kaneda A · Nucleic Acids Res 2024
    PMID 38989615DOI
    What this evidence cannot show

    Human cancer and leukaemia cell lines; the fusion status of the leukaemia cells is not stated in the abstract, so this is NOT a KMT2A-rearranged localisation result. The readout is ChIP-based occupancy at TSS - a chromatin-address measurement, not a microscope image - so it says where on the genome SETD1A sits and nothing about which nuclear compartment. Read: abstract only. Identifiers copied field by field from the efetch record fetched 2026-09-21, with the ArticleId path anchored to PubmedData/ArticleIdList/ArticleId per brief rule 81.

    Evidence that argues against this (1)
    • The domain that carries SETD1A to chromatin is the same domain that carries cyclin K, so where SETD1A is and what SETD1A does there cannot be separated by this experiment: the FLOS domain is simultaneously the BOD1L-binding chromatin tether and the cyclin-K-binding site required for chromosomal recruitment of cyclin K (this entry's E1 and E7). A FLOS mutant loses the address and the cargo at once, and no published experiment removes one without the other. Hoshii T; Cifani P; Feng Z; Huang CH; Koche R; Chen CW; Delaney CD; Lowe SW; Kentsis A; Armstrong SA 2018
  11. E_q66_loc2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset

    The reviewed human UniProt record for SETD1A (O15047, 1,707 aa, primary gene name SETD1A) carries three experimentally evidenced subcellular locations - Nucleus speckle, Chromosome and Cytoplasm - all at ECO:0000269, so the cytoplasmic pool is an experimental annotation rather than an inference by similarity.

    What this evidence cannot show

    A database aggregate: the three locations come from different papers in different cell systems, none of them haematopoietic and none KMT2A-rearranged, and UniProt does not say what fraction of the protein is in each. The speckle annotation is the weakest link to anything this graph needs, because a speckle assignment made in one cell type is routinely quoted as a property of the protein. Accession and length were read back from the fetched record and match this entry's stored uniprot and length_aa.

    Evidence that argues against this (1)
    • Reading a compartment list off a database is the move this graph has already had to undo once: localization-dot1l-nucleus-measured-cytoplasm-by-similarity exists because a cytoplasmic annotation propagated by similarity was quoted as a measurement. Here the evidence codes are experimental, but the same caution applies to the speckle row, which has no leukaemia source and is not supported by any image of SETD1A in a blood cell. UniProt Knowledgebase, fetched 2026-09-21 2026
Connected to