What this beat read and did not use: bounded absences, bounded rejections and one bounded unreadable
also called partner provenance rejections · cycle 49 enumeration · searched and dismissed
A record of the searches we ran, the papers we looked at and decided not to use, and the one paper we wanted and could not read. Without this, a future reader cannot tell what was considered from what nobody ever looked for.
The same thing, for a scientist
Per the cycle-48 instruction, this beat records all three kinds of negative. **Bounded absences.** `MLL-AF4[tiab] AND H3K79[tiab]` returns **7** records; `KMT2A-AFF1[tiab] AND H3K79[tiab]` returns **2**; `t(4;11)[tiab] AND DOT1L[tiab]` returns **3** - and that last query is unreliable because PubMed parses the parentheses and semicolon as syntax rather than text, exactly as the NUP98 beat documented. `AF9[tiab] AND AHD[tiab] AND (DOT1L[tiab] OR AF4[tiab])` returns **4**. `(AFF1[tiab] OR AF4[tiab]) AND DOT1L[tiab] AND (bind*[tiab] OR interact*[tiab])` returns **15**, of which we title-checked all 15 and judged 6 on-beat [E1]. The retired-symbol lesson applies to this project's own central fusion: the current-nomenclature query finds 2 records and the legacy one finds 7, and the decisive 2008 paper is in the legacy set only. **Bounded rejections.** Six papers or lines of work that a reader would reasonably expect to be cited here, and are not [E2][E3]. **One bounded unreadable.** PMID 37679565, the CRISPR-ChIP paper, whose full text this session could not obtain by any route; the provenance question was resolved from GEO instead, and nothing about the paper's figures or Methods is asserted anywhere in this beat [E4].
Bounded rejections: read, judged off-beat, and why
Three lines each, with what would change our minds.
PMID 37720104 - The human leukemic oncogene MLL-AF4 promotes hyperplastic growth of hematopoietic tissues in Drosophila larvae (iScience 2023). Squarely on title: human MLL-AF4, and depletion of the Drosophila DOT1L orthologue rescues the phenotype. Rejected because the recruitment question this beat asks is about a specific mammalian protein interface, and Drosophila lacks both AFF1 and the AF9/ENL AHD architecture in the form the argument turns on, so a rescue there cannot map a contact. Would change our mind: if the paper localised the requirement to a Drosophila AF4 orthologue domain with a mammalian counterpart in AFF1 2.
PMID 41614752 - a 2025 review of the KMT2A-AFF3 fusion. Returned by three of this beat's queries and about a different 3' partner; a review, not primary. Rejected as neither index fusion nor primary. Would change our mind: nothing short of it containing primary AFF-family interface data.
PMID 34174329 - The Intrinsically Disordered Proteins MLLT3 (AF9) and MLLT1 (ENL) - Multimodal Transcriptional Switches (J Mol Biol 2022). On-beat and useful as a map, and already cited once elsewhere in this graph. Not used here because this beat's rule is to cite primary structural work (PMIDs 23260655, 25921540) rather than a review of it. Would change our mind: if a claim needed the review's synthesis rather than either primary paper.
PMID 38747396 - Structural studies of intrinsically disordered MLL-fusion protein AF9 in complex with peptidomimetics (Protein Sci 2024). Directly about the AHD and about drugging it. Not used because this beat's question is which interfaces the index fusion has, not how to drug the ones it does not. Would change our mind: if the target-selection beat takes up the AHD, in which case it is the first paper to read.
PMID 30651276 - Activation of the Lysosome-Associated Membrane Protein LAMP5 by DOT1L Serves as a Bodyguard for MLL fusion oncoproteins (Clin Cancer Res 2019). Zero citations in this graph, and returned by the AFF1/DOT1L query. Not used because it concerns a downstream DOT1L target rather than recruitment; flagged here because mechanism-lamp5-as1-lncrna-activates-dot1l exists in the histone-marks beat and this paper appears to be adjacent to it and uncited 3.
PMID 26171280 - AF4 and AF4N protein complexes (Am J Blood Res 2015). Nearly rejected on journal visibility and the absence of a DOI in PubMed's record, then retained as weak evidence in mechanism-kmt2a-aff1-dot1l-recruitment-route because it is the only source this beat found asserting that AFF1-containing SECs bind DOT1L, and an entry claiming no route must confront the source claiming one. Recorded here because the near-rejection is itself the decision a future reader needs to see 3.
Bounded unreadable
PMID 37679565 / doi:10.1038/s41594-023-01087-4 - CRISPR-ChIP reveals selective regulation of H3K79me2 by Menin in MLL leukemia, Nat Struct Mol Biol 2023. What the title asserts is in the record; the abstract was read from efetch; the full text was not obtained.
Routes tried, all on 2026-09-19: Europe PMC REST (isOpenAccess N, inEPMC N, no PMCID, single full-text URL marked 'Subscription required'); Europe PMC annotations API (26 abstract-level terms only); Unpaywall (oa_status: closed, best_oa_location: null); Semantic Scholar (openAccessPdf empty, status: CLOSED); the bioRxiv preprint (10.1101/2023.03.17.533231) via its PDF, its .full HTML and its JATS source.xml, each returning HTTP 429 with Cloudflare error 1015 across four attempts and two clients.
The preprint abstract was readable via Europe PMC and is quoted where used. Nothing in this beat states what the paper probably says. The one fact the beat needed - the experimental system - was obtained from the study's GEO depositions instead, and every claim about it is attributed to GEO rather than to the paper 4. The method is written up as concept-resolving-an-unstated-system-from-geo.
A reader with institutional access should read the paper and check two things this beat could not: which figure the menin-inhibitor H3K79 result rests on, and which menin inhibitor was used.
What a re-run of this beat should do differently
- Classify every evidence item, not the first. This beat classified E1 only, on a convention. The whole-item fraction is a different and probably lower number and is worth having.
- Extend the sample to the therapeutics, resistance and target folders. They were excluded as not mechanism-bearing, and a drug-mechanism entry imports identically.
- Grep for the caveat phrase 'does not state' across the graph and run the GEO lookup on each hit. That is the work queue this beat's method implies and did not have time to execute.
- Search each topic in both nomenclatures.
MLL-AF4returns three and a half times whatKMT2A-AFF1returns on this beat's central topic.
| script | NCBI eutils esearch plus manual title-checking, 2026-09-19 |
|---|---|
| status | run |
| result summary | Seven bounded absences with exact queries, six bounded rejections with reasons and change-my-mind conditions, and one bounded unreadable (PMID 37679565) resolved by a different route. |
bounded absences — the numbers behind the chart
{
"MLL-AF4 AND H3K79": 7,
"KMT2A-AFF1 AND H3K79": 2,
"t(4;11) AND DOT1L": 3,
"AF9 AND AHD AND (DOT1L OR AF4)": 4,
"(AFF1 OR AF4) AND DOT1L AND (bind* OR interact*)": 15,
"(MLL OR KMT2A) AND ANC1 homology domain": 6,
"DOT1L AND AF10 AND octapeptide motif leucine zipper": 4
}bounded rejections
[ "37720104", "41614752", "34174329", "38747396", "30651276", "26171280 (near-rejection, retained weak)" ]
bounded unreadable
[ "37679565" ]
symbols not searched
[ "FEL", "PBM1", "ALL-1", "HRX", "Htrx" ]
- Recording rejections makes the beat auditable and also makes it look more thorough than it was: six recorded rejections out of perhaps forty abstracts skimmed means most of the reading still left no trace.
- A bounded unreadable that was worked around successfully invites the conclusion that unreadable papers do not matter; the two questions GEO could not answer - which figure, which compound - are exactly the ones that would change how much weight the partition deserves.
- Does searching AFF1's older synonyms (FEL, PBM1) and KMT2A's (ALL-1, HRX, Htrx) change any of the seven counts? This beat did not run them and the retired-symbol lesson says it should have.
- PMID 30651276 (LAMP5 activation by DOT1L) is uncited in this graph and adjacent to an existing histone-marks entry; does that beat know about it?
- Is the bioRxiv rate limit persistent or transient? A later cycle should retry 10.1101/2023.03.17.533231 before concluding the preprint is unreadable.
- E1moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
NCBI eutils esearch counts on 2026-09-19: 'MLL-AF4[tiab] AND H3K79[tiab]' 7; 'KMT2A-AFF1[tiab] AND H3K79[tiab]' 2; 't(4;11)[tiab] AND DOT1L[tiab]' 3; 'AF9[tiab] AND AHD[tiab] AND (DOT1L[tiab] OR AF4[tiab])' 4; '(AFF1[tiab] OR AF4[tiab]) AND DOT1L[tiab] AND (bind*[tiab] OR interact*[tiab])' 15; '(MLL OR KMT2A) AND ANC1 homology domain' 6; 'DOT1L AND AF10 AND octapeptide motif leucine zipper' 4.
NCBI E-utilities esearch record counts for the partner-provenance beat, run 2026-09-19 · Oncofusion knowledge graph, partner-provenance beat · NCBI E-utilities 2026What this evidence cannot show
Counts are the esearch `count` field. The 15-record query was title-checked record by record; the others were not individually title-checked except where a specific record is named in this entry. PubMed term mapping expands gene symbols to supplementary-concept records, so these are not literal string matches. The t(4;11) query is quoted as a demonstration of the parenthesis-parsing trap and should not be used as an absence claim. Wildcards were avoided except 'bind*' and 'interact*', neither of which collides with a calendar word or unit.
Evidence that argues against this (1)
- A counted absence is only as good as its symbol set. This beat searched MLL-AF4, KMT2A-AFF1, AFF1 and AF4 but did not search the older synonyms FEL or PBM1 for AFF1, nor 'ALL-1' or 'HRX' for KMT2A, so the true literature is larger than these counts by an unmeasured margin. Oncofusion knowledge graph et al. 2026
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Expression of human MLL-AF4 in the Drosophila haematopoietic system increases circulating haemocytes and enlarges the lymph gland, and depletion of Drosophila orthologues of known MLL-AF4 interactors including DOT1L, or treatment with small-molecule DOT1L inhibitors, rescues the leukaemia-like phenotype.
The human leukemic oncogene MLL-AF4 promotes hyperplastic growth of hematopoietic tissues in Drosophila larvae. · Johannessen JA, Formica M, Haukeland ALC, Bråthen NR, Al Outa A, Aarsund M, Therrien M, Enserink JM, Knævelsrud H · iScience 2023What this evidence cannot show
Abstract fetched from PubMed and read on 2026-09-19; full text not read, every identifier copied from that record. Recorded as a bounded REJECTION for this beat's purpose: Drosophila lacks AFF1 and the AF9/ENL AHD architecture the recruitment argument turns on, so a genetic rescue there cannot map a mammalian contact. It is genuinely on-topic for whether MLL-AF4 needs DOT1L, and is cited nowhere in this graph.
Evidence that argues against this (1)
- A DOT1L requirement for MLL-AF4 in an organism with no AFF1 and no AHD is arguably evidence FOR the transcription-coupled model - DOT1L is needed without any of the mammalian recruitment interfaces being present - which would make this paper support for hyp-index-fusion-h3k79-is-entirely-endogenous rather than a rejection. Biswas D et al. 2011
- E3weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper
AF4/AFF1 and AF5/AFF4 assemble super elongation complexes that recruit P-TEFb from inhibitory 7SK RNPs and bind histone methyltransferases including DOT1L, NSD1 and CARM1; PubMed's esummary record for this paper returns no DOI.
AF4 and AF4N protein complexes: recruitment of P-TEFb kinase, their interactome and potential functions. · Scholz B, Kowarz E, Rössler T, Ahmad K, Steinhilber D, Marschalek R · Am J Blood Res 2015What this evidence cannot show
Abstract fetched from PubMed and read on 2026-09-19; full text not read. Low-visibility journal, no DOI in the PubMed record, interaction network derived from overexpression and co-purification. Recorded here as a NEAR-rejection that was retained, because it is the only source this beat found asserting an AFF1-SEC to DOT1L link, and an entry claiming no mapped route must confront it. Confidence weak throughout.
Evidence that argues against this (1)
- Reconstitution from defined components finds DOT1 binds AF9 but not AF9-AF4-P-TEFb complexes, the direct opposite of an SEC carrying DOT1L; the two reports are in tension and neither maps a contact on AFF1. Biswas D et al. 2011
- E4established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper
The full text of PMID 37679565 could not be obtained in this session by any of five routes - Europe PMC, Europe PMC annotations, Unpaywall, Semantic Scholar, or the bioRxiv preprint's PDF, HTML and JATS XML - and the experimental system was instead established from the study's GEO depositions.
CRISPR-ChIP reveals selective regulation of H3K79me2 by Menin in MLL leukemia. · Gilan O, Talarmain L, Bell CC, Neville D, Knezevic K, Ferguson DT, Boudes M, Chan YC, Davidovich C, Lam EYN, Dawson MA · Nat Struct Mol Biol 2023What this evidence cannot show
All routes attempted on 2026-09-19 from one network location. HTTP 429 is a rate limit rather than a permission denial, so the bioRxiv preprint is likely readable from a different network or at a later time; this is a bounded unreadable for this session, not a permanent one. Institutional subscription access was not available to this agent.
Evidence that argues against this (1)
- The preprint abstract WAS readable through Europe PMC and states the synergy rationale more explicitly than the published abstract, so the paper is not wholly unreadable - only its figures and Methods are, which is exactly where the two remaining questions live. Gilan O et al. 2023
the beat's negative record
adds one bounded unreadable
the route used when the paper could not be read
the nomenclature asymmetry that hid the index-fusion literature