No published experiment combines an AKAP95 perturbation with any menin or DOT1L inhibitor - bounded by an enumerated 92-record literature, not by a keyword zero
also called AKAP95 menin inhibitor absence
AKAP95 is now a known dependency of MLL-rearranged leukaemia, and menin inhibitors are the approved drugs for that disease. Nobody has yet put the two together in a single experiment.
The same thing, for a scientist
**The claim, at the strongest level the instruments support.** The complete AKAP8/AKAP95 title-and-abstract literature is **92 PubMed records**; all 92 were fetched and scanned, and **none** combines AKAP95 with a menin inhibitor, a DOT1L inhibitor, or any KMT2A-directed agent [E1]. The two full texts that contain the disease biology - PMID 23995757 and PMID 41501053 - were read end to end, and **neither contains the string revumenib, SNDX-5613, VTP-50469, ziftomenib, KO-539, pinometostat, EPZ-5676, MI-503 or menin** [E2][E3]. The abstract-level drug-name panel returns 0 for every one of those names, but **that zero is worthless on its own**, because every cell-line token in the same panel also returns 0 in a literature that contains published knockdown curves in three of those lines [E1]. **The bound that carries weight is the enumeration plus the two full texts, and it should be quoted that way.**
Why this gap is worth closing, and what the prediction is
This is not a gap for completeness. The two mechanisms make opposite predictions and the experiment is one plate.
AKAP95 reaches the KMT2A axis by co-condensing with the retained N-terminal fragment of KMT2A (interface-akap8-kmt2a-n-terminal-fragment). A menin inhibitor displaces the fusion from chromatin. If the AKAP95 requirement is downstream of fusion chromatin occupancy, then under revumenib the fusion is no longer on the genes, the co-condensate has no anchor, and AKAP95 should become dispensable - the two agents would be epistatic, not additive, and combining them would buy nothing.
If instead they are additive or synergistic, AKAP95 is doing something the fusion's chromatin residence does not account for, and the combination is a real therapeutic proposition in a disease where menin-inhibitor resistance is already documented.
Either result is publishable and neither is predictable from what exists. Adding a revumenib arm to the rescue experiment in exp-separate-the-dpy30-route-from-the-mll1t-route costs one plate.
And the honest deflation: a negative - AKAP95 loss adds nothing under menin inhibition - would substantially reduce AKAP95's therapeutic interest, because the population that needs a new agent is precisely the menin-inhibitor-treated one.
The instrument note, because it is the transferable part
The drug-name panel and the cell-line panel returned identical answers - all zeros - and one of them was right and one was wrong.
| panel | abstract-level result | truth |
|---|---|---|
| cell lines (MV4-11, MOLM-13, RS4;11, SEM, THP-1, NOMO-1) | 6/6 zero | FALSE - three carry published AKAP95 knockdown curves |
| drugs (revumenib group, ziftomenib group, pinometostat group, MI-503 group, menin) | 5/5 zero | TRUE, as far as full text can establish |
A panel of zeros cannot tell you which of its rows is a false zero. The project's agents-and-models rule says to query drug and cell-line names rather than disease terms, and it is right - but at abstract level both classes live in the Methods, and the method that distinguishes them is reading the papers. Here that was affordable because the literature is 92 records.
Rule of thumb this beat would add: when every row of an agents-and-models panel returns zero, treat the panel as uninformative and escalate, rather than reading uniformity as corroboration.
| script | analyses/q166-the-akap95-dpy30-axis/ (akap-corpus-92-titles.json, pubmed_nomen.json) |
|---|---|
| status | run |
| result summary | No published experiment combines an AKAP95 perturbation with any menin or DOT1L inhibitor. Bounded by a complete 92-record title-abstract enumeration plus two full texts read end to end, not by a keyword zero - and the accompanying cell-line panel demonstrates why the keyword zero alone would have been worthless. |
| absence class | never-measured |
inputs
[ "PubMed esearch/efetch via scripts/pubmed_safe.py", "full texts of PMC3813012 and PMC12886973" ]
names tried
[ "revumenib", "SNDX-5613", "VTP-50469", "ziftomenib", "KO-539", "pinometostat", "EPZ-5676", "EPZ5676", "MI-503", "MI503", "MI-2-2", "menin" ]
measurements — the numbers behind the chart
[
{
"quantity": "count",
"subject": "records in the complete AKAP8/AKAP95 literature naming any menin or DOT1L inhibitor",
"value": 0,
"units": "records",
"n": 92,
"universe": "all 92 PubMed records matching AKAP8[tiab] OR AKAP95[tiab], efetched with abstracts and scanned, 2026-09-24",
"null": null,
"note": "rule-of-three upper bound: under 3.3% of this literature, i.e. the true count is almost certainly 0 or 1"
},
{
"quantity": "count",
"subject": "drug-name tokens tried in the abstract-level panel",
"value": 12,
"units": "tokens",
"n": 12,
"universe": "revumenib, SNDX-5613, VTP-50469, ziftomenib, KO-539, pinometostat, EPZ-5676, EPZ5676, MI-503, MI503, MI-2-2, menin; hyphenated and unhyphenated compound codes treated as distinct strings per rule 72",
"null": null,
"note": "all returned 0; the translated list is published in pubmed_nomen.json so a reader can judge the absence"
},
{
"quantity": "count",
"subject": "cell-line tokens in the same panel that returned a FALSE zero",
"value": 3,
"units": "tokens",
"n": 6,
"universe": "MV4-11, MOLM-13, THP-1 all have published AKAP95 knockdown growth curves and all return 0 at abstract level",
"null": null,
"note": "THE CONTROL THAT MATTERS. Half the cell-line panel is demonstrably false, which is why the drug-panel zeros are quoted only alongside the full-text reads."
}
]- Stating this absence invites the combination experiment, and the most likely result is the deflating one: if AKAP95's requirement sits downstream of fusion chromatin occupancy, menin inhibition already removes it and the combination adds nothing in exactly the patients who need a second agent.
- A bounded absence is an invitation to be the first, which is a poor reason to run an experiment. The argument for running this one is the epistasis question, not the novelty.
- Is AKAP95 loss epistatic to menin inhibition, or additive? One plate answers it, and the two outcomes point in opposite therapeutic directions.
- Is AKAP95 among the proteins whose level or localisation changes under revumenib in a KMT2A-rearranged line? Any published menin-inhibitor proteomics dataset could be re-mined for this without new wet work, and none has been.
- Does AKAP95 loss sensitise or protect against menin-inhibitor-driven differentiation? If AKAP95 is required to execute state transitions, it should PROTECT - the prediction that would most damage the therapeutic story, and the cheapest one to test.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.dataset
All 92 PubMed records matching AKAP8[tiab] OR AKAP95[tiab] were fetched with titles and abstracts and scanned; none names a menin inhibitor, a DOT1L inhibitor or any KMT2A-directed agent. At abstract level the paired panel returns 0 for (AKAP8 OR AKAP95) with each of revumenib/SNDX-5613/VTP-50469, ziftomenib/KO-539, pinometostat/EPZ-5676/EPZ5676, MI-503/MI503/MI-2-2, and menin - and also 0 for each of MV4-11/MV4;11/MV411, MOLM-13/MOLM13, RS4;11/RS4-11, SEM AND leukemia, THP-1/THP1 and NOMO-1/NOMO1, three of which are cell lines with published AKAP95 knockdown growth curves. Bare-operand controls confirm the operands are alive: AKAP8 30, AKAP95 66, union 92, KMT2A 1654.
analyses/q166-the-akap95-dpy30-axis/ - akap-corpus-92-titles.json (complete efetch) and pubmed_nomen.json (29 queries with querytranslation), 2026-09-24, via scripts/pubmed_safe.py · this project 2026What this evidence cannot show
A title-and-abstract enumeration cannot see a menin-inhibitor experiment reported in a supplementary table of a paper that is not about AKAP95 - the same rule-80 blind spot that produced the original false absence about AKAP95 in leukaemia. This bound is therefore strong but not exhaustive, and it is strongest for the two papers whose full texts were read.
Evidence that argues against this (1)
- The enumeration is of AKAP8/AKAP95's own literature. A menin-inhibitor paper that profiled AKAP95 among thousands of proteins - a proteomics or CRISPR screen under revumenib - would not name AKAP95 in its title or abstract and would be invisible here. The MLL-fusion interactome that first flagged AKAP95 is exactly such a paper and it is outside this 92-record set. Yan X et al. 2026
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The full text of PMID 41501053, the only paper reporting AKAP95 in MLL-rearranged leukaemia, contains no menin inhibitor, no DOT1L inhibitor and no menin experiment of any kind. Its therapeutic arm is the JD-PI95 peptide alone, with no combination.
AKAP95 condensates regulate transcription and can be targeted in MLL-fusion driven oncogenesis. · Yan X, Luan Y, Zhu Y, Cai Y, Ye F, Yang L, Zhu K, Hu J, Luo S, Lin J, Wang Y, Huang Y, Sun P, Yang Z, Shi B, Jiang H, et al. · Nat Commun 2026What this evidence cannot show
Full text read via PMC12886973; supplementary figures and Source Data were NOT available, so a combination experiment buried in a supplementary panel would have been missed. The expected_absent_terms are the support: the absence of those strings is the claim.
Evidence that argues against this (1)
- This is a 2026 paper, so the absence is current rather than historical, and a menin-inhibitor combination is the single most obvious follow-up. The gap may already be closed in work not yet indexed, and a beat quoting this absence in a later cycle must re-run the enumeration rather than quote the stored count. this project 2026
- E3established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The full text of PMID 23995757, the founding AKAP95-MLL paper, contains no leukaemia cell, no fusion and no inhibitor of any kind; its systems are Flp-In-293, HeLa nuclear extract, G-293T, NT2 and mouse E14 ESCs.
Regulation of transcription by the MLL2 complex and MLL complex-associated AKAP95. · Jiang H, Lu X, Shimada M, Dou Y, Tang Z, Roeder RG · Nat Struct Mol Biol 2013What this evidence cannot show
Full text read. Note that MENIN itself does appear in this paper - as a subunit identified by mass spectrometry in the purified FH-MLL2 complex (Fig 1A) and as a commercial antibody. What is absent is any menin INHIBITOR or menin experiment; the expected_absent_terms list names the inhibitor, not the protein, for that reason.
Evidence that argues against this (1)
- Menin inhibitors did not exist in publishable form in 2013, so this paper's silence on them carries no information about the field's interest. It is recorded only to establish that the founding paper contributes nothing to the combination question, not as evidence that nobody wanted to ask. Jiang H et al. 2013
the bounded absence: 0 of 92 enumerated records, plus two full texts read end to end
never combined with any AKAP95 perturbation; the epistasis question is one plate and points in opposite therapeutic directions
- related to →Pinometostat (EPZ-5676) — the DOT1L inhibitor, and the most instructive failure in this field
the DOT1L arm of the same absence; pinometostat, EPZ-5676 and EPZ5676 all return zero
uses the same enumeration, and inherits its cell-line panel as the control showing why a keyword zero alone is worthless here