Oncofusion

start / graph / gene-ythdf2

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

YTHDF2 — the cytoplasmic m6A reader that selectively compromises AML cancer stem cells, discordant across the lineage split here

also called HGRG8 · NY-REN-2 · YTH domain family member 2 · high-glucose-regulated gene 8

YTHDF2 reads the RNA tag in the cytoplasm and sends the tagged messages to be destroyed. Removing it damages leukaemia stem cells while sparing normal blood stem cells, which is the shape a good target has.

The same thing, for a scientist

YTHDF2 is the m6A reader that directs bound transcripts to decay. Targeting it selectively compromises cancer stem cells in acute myeloid leukaemia while sparing normal haematopoietic stem cells [E1], the clearest therapeutic-window claim in the m6A field. This project's scan finds it **lineage-discordant**: -0.125 in four B-lymphoid KMT2A-rearranged lines and +0.065 in seven myeloid ones, Chronos mean -0.343 with 27.6% of 1,208 lines depleted, at the 49.1st null percentile [E2]. Against MLL or KMT2A the whole YTHDF family returns 3 records [E3].

Go deeper
The window claim, and what it rests on

The reason YTHDF2 matters to this graph is not its mechanism but its selectivity claim: cancer stem cells compromised, normal haematopoietic stem cells spared 1. That is the profile every target in this beat is being assessed against, and very few have it.

Two cautions attach. The sparing of normal HSCs is measured in a mouse in a defined assay, and this graph's menin work is the standing example of a normal-tissue result that held at steady state and failed under competitive transplantation. And the disease is AML; there is no B-lymphoid arm, and this project's index-arm data are discordant across the lineage split, with the myeloid half - where the paper is - showing a positive difference 2.

Discordance again, and the pattern it forms

YTHDF2 joins METTL3, MSI2, IGF2BP1 and ALKBH5 as genes whose two lineage halves disagree in sign in the index arm 2. Thirty of 79 panel genes are discordant.

That is a high rate, and it is itself a finding about the instrument rather than about any one gene: with four lymphoid and seven myeloid lines and a null SD of 0.075, sign agreement between halves is not the default. It means any single-gene claim from this scan - including the positive ones - needs the concordance check that ZC3H13, SYNCRIP, HNRNPA2B1 and PTBP1 pass and these five fail.

Facts and figures
uniprotQ9Y5A9
length AA579
gene symbolYTHDF2
uniprot resolutiongene_exact:YTHDF2 AND organism_id:9606 AND reviewed:true; one hit; primary gene name read back as YTHDF2; length 579 aa checked
rna functioncytoplasmic m6A reader directing bound transcripts to decay
paraloguesYTHDF1 (Chronos mean +0.079, 0.0% depleted) and YTHDF3, widely argued to be functionally redundant
essentiality
{
  "depmap_common_essential": false,
  "chronos_mean": -0.3425,
  "pct_lines_depleted": 27.6,
  "n_lines": 1208,
  "kmt2a_r_lymphoid_diff": -0.1253,
  "kmt2a_r_myeloid_diff": 0.0645,
  "null_percentile": 49.14,
  "lineage_discordant": true
}
pubmed counts
{
  "YTHDF2 AND leukemia": 28,
  "(MLL OR KMT2A) AND (YTHDF2 OR YTHDF1 OR YTHDF3)": 3
}
depmap kmt2a selectivity
{
  "chronos_mean_kmt2a_rearranged": -0.4354,
  "chronos_mean_other_leukaemia": -0.4309,
  "difference": -0.0045,
  "pan_essentiality_all_lines": -0.3425,
  "selectivity_rank": 8557,
  "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
  • The three YTHDF paralogues are argued to be redundant, so a selective YTHDF2 agent may be compensated immediately.
  • The normal-HSC sparing is a mouse steady-state result; this graph has a worked example of such a result failing under competitive stress.
  • The scan's myeloid half - where the supporting paper is - shows no extra requirement in KMT2A-rearranged lines, so the fusion-specific case is unmade even within AML.
Still open
  1. Does the AML cancer-stem-cell window hold in B-lymphoid KMT2A-rearranged disease, where nobody has looked?
  2. Do YTHDF1/2/3 buffer one another, which would make every single-gene result in this family an under-report?
  3. Does the normal-HSC sparing survive competitive transplantation, the condition under which this graph's menin result changed?
Every source behind this entry (4)
  1. E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Targeting the RNA m6A reader YTHDF2 selectively compromises cancer stem cells in acute myeloid leukaemia.

    PMID 31031138DOI
    What this evidence cannot show

    Abstract-level. AML; no B-lymphoid or KMT2A::AFF1 arm. The sparing of normal haematopoietic stem cells is measured in murine assays under conditions that this graph has repeatedly found to be the ones least likely to reveal a self-renewal toxicity.

    Evidence that argues against this (1)
    • This project's scan gives YTHDF2 a positive index-arm difference (+0.065) in the seven myeloid KMT2A-rearranged lines - the lineage where the paper is - against -0.125 in the four B-lymphoid ones. The two halves disagree in sign, so the scan supplies no support for the published direction. Oncofusion knowledge graph (internal analysis) 2026
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.analysis

    DepMap CRISPR (Chronos) gives YTHDF2 a mean gene effect of -0.343 across 1,208 lines with 27.6% depleted below -0.5, and lineage-DISCORDANT index-arm differences of -0.125 (four B-lymphoid) and +0.065 (seven myeloid) against 43 other leukaemia lines, at the 49.1st percentile of a 466-gene null. Thirty of 79 panel genes are lineage-discordant.

    What this evidence cannot show

    Four and seven index lines. A discordant gene's pooled number is a lineage average of contradictory results and must not be quoted.

    Evidence that argues against this (1)
    • A cancer-stem-cell phenotype is exactly what a two-to-three-week bulk proliferation dropout cannot measure, so a null or discordant Chronos row is weak evidence against the published claim rather than evidence for the opposite. Paris J et al. 2019
  3. E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database

    Counted: `(MLL OR KMT2A) AND (YTHDF2 OR YTHDF1 OR YTHDF3)` returns 3 PubMed records for the whole reader family, against 28 for `YTHDF2 AND leukemia`.

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

    Family-wide query run to avoid partitioning a redundant-reader literature across three symbols. Three records is a literature too small for any count to be robust.

    Evidence that argues against this (1)
    • The three YTHDF paralogues are widely argued to act redundantly on the same m6A-marked transcripts, so single-gene counts and single-gene dependency rows both under-report the family - the same buffering problem as the IGF2BP and KHDRBS families in this beat. 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 Q9Y5A9 (YTHD2_HUMAN) is human YTHDF2, 579 amino acids; resolved by gene_exact query with the primary gene name read back as YTHDF2 and the length checked.

    What this evidence cannot show

    Single reviewed human hit whose primary gene name equals the query symbol. This project has logged six cases of gene_exact returning a protein that merely lists the queried symbol as a synonym, so the read-back is the point, not the query.

    Evidence that argues against this (1)
    • An accession establishes identity and length. It says nothing about whether this protein matters in KMT2A-rearranged leukaemia, which is what the rest of this entry is for. UniProt Consortium 2026
Connected to