Oncofusion

start / graph / gene-mettl14

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q6-interactome-rna-layer

METTL14 — the other half of the m6A writer, with a real AML mechanism and zero KMT2A-specific records

also called MT-A70-like · methyltransferase-like 14 · methyltransferase 14, N6-adenosine-methyltransferase subunit

METTL14 is the partner enzyme that works with METTL3 to tag messenger RNA. It blocks blood cells from maturing and promotes myeloid leukaemia, but nobody has published on it in KMT2A-rearranged disease specifically.

The same thing, for a scientist

METTL14 forms the catalytic heterodimer with METTL3. It inhibits haematopoietic stem/progenitor differentiation and promotes leukaemogenesis via mRNA m6A modification [E1]. Against this graph's index disease the literature is empty at the symbol level: `(MLL OR KMT2A) AND METTL14` returns one record, which title-checking shows to be a review of bivalent chromatin that mentions METTL14 among novel regulators and is not a leukaemia study [E2]. DepMap gives it a Chronos mean of -0.716 with 78.8% of 1,208 lines depleted and an index-arm difference of -0.016 B-lymphoid / -0.188 myeloid [E3] - broadly required, mildly myeloid-leaning, not selective.

Go deeper
A one-record count that is really a zero

This is a worked example of why raw counts are not claims. (MLL OR KMT2A) AND METTL14 returns 1 record. Title-checked, that record is a review of bivalent chromatin in development and cancer which lists METTL14 among emerging regulators of bivalent domains alongside DPPA2/4, QSER1, BEND3 and TET1, and which mentions MLL/COMPASS as the H3K4me3 depositor 2. The two terms co-occur; the paper is about neither in combination.

So the honest statement is zero KMT2A-specific METTL14 studies, and the raw count of one would have looked like a thin literature rather than an absent one. This project logged the opposite failure - alias inflation turning 2 real records into 115 for NELFE - and this is the small-count version: three of seven ICE1/ICE2 records were incidental mentions, and here one of one is.

Where it is real, and why that does not transfer

METTL14 inhibits haematopoietic stem/progenitor differentiation and promotes leukaemogenesis through mRNA m6A 1, which is a genuine haematopoietic result. It is also myeloid, and it belongs to the same body of work as the METTL3 story - same complex, same modification, same lineage.

The complex-level point matters for enumeration. METTL3 and METTL14 function as an obligate heterodimer, so a subunit-keyed search partitions one literature across two symbols, and a paper framing its result as 'the m6A methyltransferase complex' is reachable from neither. That phrase returns one leukaemia record in PubMed 2. This is the NSL-complex trap from cycle 45 in a different guise, and it is the reason both subunits get entries here rather than one.

Facts and figures
uniprotQ9HCE5
length AA456
gene symbolMETTL14
uniprot resolutiongene_exact:METTL14 AND organism_id:9606 AND reviewed:true; one hit; primary gene name read back as METTL14; length 456 aa checked
rna functionRNA-binding, catalytically inactive partner of METTL3 in the m6A methyltransferase complex
essentiality
{
  "depmap_common_essential": false,
  "chronos_mean": -0.7164,
  "pct_lines_depleted": 78.8,
  "n_lines": 1208,
  "kmt2a_r_lymphoid_diff": -0.016,
  "kmt2a_r_myeloid_diff": -0.1881,
  "null_percentile": 5.15
}
pubmed counts
{
  "METTL14 AND leukemia": 58,
  "(MLL OR KMT2A) AND METTL14": "1 raw, 0 after title check"
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.9472,
  "chronos_mean_other_leukaemia": -0.8217,
  "difference": -0.1255,
  "pan_essentiality_all_lines": -0.7164,
  "selectivity_rank": 709,
  "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
  • Depleted in 78.8% of DepMap lines: an essential-function toxicity risk before any disease-specific argument is made.
  • The subunit is catalytically inactive, so the existing chemical matter against METTL3 does not address it and a separate modality would be needed if the two turn out to be separably required.
Still open
  1. Does the METTL3-METTL14 heterodimer matter in KMT2A::AFF1 B-lymphoid cells at all, a question with zero published records?
  2. Are the two subunits separable as targets, given that only METTL3 carries the catalytic site that STM2457 occupies?
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    METTL14 inhibits haematopoietic stem/progenitor cell differentiation and promotes leukaemogenesis via mRNA m6A modification.

    PMID 29290617DOI
    What this evidence cannot show

    Abstract-level. Myeloid haematopoiesis and myeloid leukaemia; no KMT2A::AFF1 or B-lymphoid arm.

    Evidence that argues against this (1)
    • DepMap gives METTL14 a B-lymphoid KMT2A-rearranged index-arm difference of -0.016, which is indistinguishable from zero against a null of SD 0.075. Whatever METTL14 does in myeloid leukaemogenesis is not visible as a B-lymphoid fusion-selective dependency. Oncofusion knowledge graph (internal analysis) 2026
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    Counted absence: `(MLL OR KMT2A) AND METTL14` returns 1 PubMed record, which on title check is a review of bivalent chromatin in development and cancer that names METTL14 among emerging regulators and mentions MLL/COMPASS separately - not a study of METTL14 in KMT2A-rearranged leukaemia. `"m6A methyltransferase complex" AND leukemia` returns 1 record, which is not KMT2A-specific either.

    PubMed E-utilities esearch counts, alias-expanded and title-checked · NCBI PubMed (queries run 2026-09-20) 2026
    What this evidence cannot show

    Symbol-keyed counts partition an obligate-heterodimer literature across two subunits and miss any paper framed at complex level, which is why the complex phrase was searched separately. It returns one record.

    Evidence that argues against this (1)
    • The relevant paper was found by title-checking rather than by counting, which is the point: the count of one looks like a thin literature and is an absent one. Any counted-absence claim in this beat that was not title-checked should be read as provisional. NCBI PubMed (queries run 2026-09-20) 2026
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.analysis

    DepMap CRISPR (Chronos) gives METTL14 a mean gene effect of -0.716 across 1,208 lines with 78.8% depleted below -0.5, and an index-arm difference of -0.016 (four B-lymphoid) and -0.188 (seven myeloid) against 43 other leukaemia lines, at the 5.2nd percentile of a 466-gene null.

    What this evidence cannot show

    Near-broadly-required, so the dynamic range for selectivity is small. Four and seven index lines.

    Evidence that argues against this (1)
    • A 5.2nd-percentile combined difference sounds selective and is not: the number is driven by the myeloid half, the effect size is -0.188 Chronos units, and the percentile is against a null of 466 genes whose SD is 0.075. Percentile without effect size is the error this project corrected in cycle 45. Oncofusion knowledge graph (internal analysis) 2026
  4. E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    UniProtKB Q9HCE5 (MET14_HUMAN) is human METTL14, 456 amino acids; resolved by gene_exact query with the primary gene name read back as METTL14 and the length checked.

    What this evidence cannot show

    Single reviewed human hit with matching primary name. METTL14 is catalytically inactive on its own and functions as the RNA-binding partner of METTL3, so an accession-level statement about 'the methyltransferase' is ambiguous between subunit and complex.

    Evidence that argues against this (1)
    • Identity only. That METTL14 is the non-catalytic partner means an inhibitor strategy aimed at the METTL3 active site does not target it, and any claim of redundancy or of separate function between the two subunits needs evidence the accession does not provide. Yankova E et al. 2021
Connected to