Enumeration for the index-fusion model beat: what was searched, found, rejected and could not be read
also called Q21 enumeration
A record of everything this piece of work looked at, including the things it decided not to use and why, so that someone repeating it next year repeats the decision and not just the search.
The same thing, for a scientist
The beat's audit trail. **Searched:** Cellosaurus by free-text fusion term, by translocation string, by line identifier and by DepMap cross-reference; PubMed by capability, resource and line-name queries; the Open Targets DepMap essentiality endpoint; Europe PMC and `efetch db=pmc` for full texts. **Found and written up:** 11 lines and panels, 4 engineered model classes, 3 xenograft resources, 1 bounded absence, 1 verdict. **Rejected with reasons:** 9 lines and 4 paper classes. **Could not read:** 1 paper with a PMCID whose full text neither route returned.
Bounded rejections — KMT2A::AFF1 lines enumerated and not written up
Each was resolved by Cellosaurus accession on 2026-09-19 and set aside for the stated reason. Each would be worth an entry if the beat's question changed.
| line | accession | age/sex | reason for rejection | what would change my mind |
|---|---|---|---|---|
| SEM-K2 / SEMK2-M1 | CVCL_S906 / CVCL_S907 | F 5Y | subclone of SEM, same individual; covered inside the SEM entry | evidence the subclone's dependencies differ from the parent's |
| EU-8 | CVCL_8866 | F 1Y | AG 1Y is ambiguous between 12 and 23 months, so infant status is unverifiable from the record | a founding report giving age in months |
| AN4;11 | CVCL_AY52 | M 8M | infant and in-frame KMT2A-AFF1, but annotated childhood T-ALL and established from a tumour grown in SCID mouse peritoneum, so it is neither B-lineage nor a direct patient derivation | confirmation that the donor's disease was genuinely T-lineage — a t(4;11) T-ALL in an 8-month-old would deserve its own entry and possibly its own beat |
| ALL-2 / ALL-2B | CVCL_QW62 / CVCL_QW67 | F 10Y | not infant; ALL-2's fusion annotation is inherited 'from child cell line ALL-2B' rather than measured in ALL-2 | a need for more non-infant comparators |
| B1 [Human leukemia] | CVCL_QW66 | M 14Y | not infant | as above |
| BEL-1 | CVCL_0081 | F 41Y | adult | a beat on adult KMT2A::AFF1 |
| Kid92 | CVCL_A088 | M 28Y | adult, and Cellosaurus carries a Caution that the founding paper says 16 years in its text and 28 in its Table 1 | nothing for this beat; it is a good worked example of a depositor-age discrepancy |
| CNLBC1 | CVCL_V650 | F 63Y | returned by the t(4;11)(q21;q23) query but is a chronic neutrophilic leukaemia line with a septin-family KMT2A partner, not AFF1 | nothing — it is a query artefact, recorded so nobody re-investigates |
| RS4;11 BCL2-F104L / BCL2-G101V / mouse MCL1; MV4-11 TP53-KO / CEBPA-KO / MV-4-11B | CVCL_XZ61, XZ62, E8JJ, F4CE, F1S5, C3HG | — | engineered derivatives of lines already covered; they inflate any count of 'KMT2A::AFF1 lines' and represent no additional patient | a question about venetoclax resistance, where the BCL2-mutant RS4;11 derivatives are the point |
Bounded rejections — papers and resources that look on-beat and are not
- PMID 37494611, Establishment of a t(11;19), KMT2A Rearranged B-ALL Cell Line for Preclinical Evaluation. A newly established KMT2A-rearranged B-ALL line, 2023 — exactly what this beat was looking for, with the wrong partner (MLLT1, not AFF1). Would be central to a KMT2A::MLLT1 beat.
- PMID 27317766, CCI-007, a novel small molecule with cytotoxic activity against infant leukemia with MLL rearrangements. The only record in PubMed whose title or abstract names any Perth PER line. It is a compound paper, not a model paper, so it is cited here only as evidence of how poorly cell-line names are indexed.
- PMID 14999297, FLJ10849, a septin family gene, fuses MLL in a novel leukemia cell line CNLBC1. The origin of the CNLBC1 artefact above.
- The PPTC supplementary Table S5. Not rejected — not retrieved. It holds the KMT2A partner composition and donor ages of the PPTC MLL-ALL PDX group and is the single highest-value unretrieved item in this beat. Recorded as a bounded unresolved in
panel-pptc-childhood-pdx-and-the-mll-all-group. - Drug-efficacy literature in the Perth panel (romidepsin, carfilzomib, dactinomycin, azacitidine/venetoclax, irinotecan, ONX-0914, ribavirin, auranofin, CBL0137/panobinostat, HDAC7). Enumerated and deliberately not written up: this beat is an inventory of material, and each of these belongs to the therapeutics beats. They are cited here only where they establish that a model exists and is used.
Bounded unreadable
One paper meets the project's threshold — a PMCID exists and both fullTextXML and efetch db=pmc failed to return a <body>:
- PMID 22212479, A human ESC model for MLL-AF4 leukemic fusion gene reveals an impaired early hematopoietic-endothelial specification, Cell Research 2012, PMC3367544. Europe PMC
fullTextXMLand NCBIefetch db=pmcboth failed to yield a body element on 2026-09-19. The abstract was read and is the basis of the claims made from it; nothing beyond the abstract has been written about this paper.
Every other key paper in the beat was readable and the readability was tested rather than assumed. Full-text bodies confirmed present for PMIDs 33937032 (PMC8082024), 39941894 (PMC11816686), 32414848 (PMC8018117), 36380143 (PMC9883157), 34824279 (PMC8616957), 27396339 (PMC4967476), 27666791 (PMC5063541) and 31693904 (PMC6880934, recovered via efetch db=pmc after Europe PMC returned HTTP 500). No PMCID exists for PMIDs 27846391, 16607274, 23479570 or 37639801.
Reading two of those full texts changed conclusions, which is why the list is worth keeping: PMC8082024 gave the ages in days that overturn the standing weak spot, and PMC8018117 revealed that a KMT2A-AFF1 model in an 'infant ALL PDX' panel is adult.
Searched and found nothing
Recorded in full, with counts, in concept-no-dependency-screen-in-infant-kmt2a-aff1-material. In short: two queries return zero — CRISPR/dependency screening crossed with infant-leukaemia terms, and CRISPR crossed with the infant line names.
One query, C, returned 41 records dominated by parser artefacts (SARS-CoV-2 spike protein, animal genomic kinship, coriander powder in fish feed, pulse oximeter bench tests), because PubMed splits hyphenated short tokens such as ALL-PO and BLIN-3. That number is reported here only as a demonstration that cell-line names are not reliably searchable in PubMed, which is the methodological finding of this section and applies to any future beat trying to count how much a line has been used.
| script | manual enumeration; no new script |
|---|---|
| status | run |
| run date | 2026-09-19 |
| written up | 18 |
| rejected lines | 9 |
| rejected paper classes | 4 |
searched
[ "Cellosaurus search API, q=KMT2A-AFF1 (29 records)", "Cellosaurus search API, q=t(4;11)(q21;q23) (5 records)", "Cellosaurus search API, q=MLL-AF4 (31 records)", "Cellosaurus search API, q=id:<PER line> for eight Perth lines", "Cellosaurus search API, q=dr:<ACH id> for the eleven DepMap screened lines", "Cellosaurus cell-line API by accession for 24 accessions", "Open Targets GraphQL depMapEssentiality screens for MEN1 (1,258 lines)", "PubMed esearch, five stated queries on screening, line names and PDX", "PubMed esearch on model-establishment, mouse genetics, hESC and iPSC routes", "Europe PMC core + fullTextXML and efetch db=pmc for eleven papers" ]
unretrieved high value
[ "PPTC supplementary Table S5 (KMT2A partner composition and donor ages of the MLL-ALL PDX group)" ]
unreadable
[
{
"pmid": "22212479",
"pmcid": "PMC3367544",
"routes_tried": [
"Europe PMC fullTextXML",
"efetch db=pmc"
],
"outcome": "no body element on 2026-09-19",
"title": "A human ESC model for MLL-AF4 leukemic fusion gene reveals an impaired early hematopoietic-endothelial specification.",
"authors": "Bueno C, Montes R, Melen GJ, et al.",
"venue": "Cell Res",
"year": 2012,
"title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}
]full text read
[ "33937032 (PMC8082024)", "31693904 (PMC6880934)", "32414848 (PMC8018117)" ]
methodological findings
[ "A fusion-keyed Cellosaurus query cannot see the Perth infant lines, which carry no gene-fusion annotation.", "Cellosaurus age and disease-term fields can point different ways within one record (ALL-PO: AG 4M, DI childhood B-ALL).", "PubMed splits hyphenated short cell-line tokens, so a line-name query returns unrelated records and cannot be counted.", "An abstract describing a panel as 'infant' does not mean every model in it is; PMC8018117 names one KMT2A-AFF1 model as adult." ]
- A rejection list is only as good as the question it was written against. Every line rejected here is rejected relative to 'infant-derived human KMT2A::AFF1 material for dependency work', and a beat with a different question should re-read the reasons rather than the conclusions.
- Two of this beat's conclusions changed on reading a full text. Any entry in this beat whose caveat says 'abstract read; full text not read' should be treated as provisional in the same way, and several say exactly that.
- Is AN4;11 (CVCL_AY52) genuinely a T-lineage leukaemia from an 8-month-old with in-frame KMT2A::AFF1? If so it is a counterexample to the fusion's B-lineage instructiveness and this beat rejected it too quickly.
- Does PPTC Table S5 contain infant KMT2A::AFF1 PDX models? One download.
- Would a full-text search engine find dependency screening in infant lines that a [tiab] query cannot see?
- E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database
Nine KMT2A::AFF1-annotated cell lines were resolved by Cellosaurus accession and set aside from this beat with stated reasons: SEMK2 (CVCL_S906) and SEMK2-M1 (CVCL_S907) as SEM subclones, EU-8 (CVCL_8866) for an ambiguous 1Y age, AN4;11 (CVCL_AY52) as an 8-month-old's childhood T-ALL established through SCID-mouse passage, ALL-2 (CVCL_QW62) and ALL-2B (CVCL_QW67) at 10 years, B1 (CVCL_QW66) at 14 years, BEL-1 (CVCL_0081) at 41 years, and Kid92 (CVCL_A088) at 28 years with a recorded age discrepancy between its founding paper's text and its Table 1.
Cellosaurus records for CVCL_S906, CVCL_S907, CVCL_8866, CVCL_AY52, CVCL_QW62, CVCL_QW67, CVCL_QW66, CVCL_0081 and CVCL_A088 · Bairoch A, et al. (SIB Swiss Institute of Bioinformatics) · Cellosaurus 2025What this evidence cannot show
All fetched by accession on 2026-09-19 with primary names read back. Rejection is relative to this beat's question (infant-derived human KMT2A::AFF1 material for dependency work) and is not a judgement of the lines' quality. AN4;11 in particular is a genuinely infant-derived in-frame KMT2A-AFF1 line and its exclusion rests on the T-lineage disease term and the xenograft-passage derivation, both of which a different beat might weigh differently.
Evidence that argues against this (1)
- Excluding AN4;11 may be the wrong call. An 8-month-old with an in-frame KMT2A::AFF1 and a T-lineage phenotype is a counterexample to the claim that this fusion produces B-lineage disease, which is a claim the mouse genetics of this fusion is specifically cited for. If the T-ALL assignment is correct it belongs in the lineage beat, not in a rejection list. Metzler M et al. 2006
- E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
A 2023 report establishes a new KMT2A-rearranged B-ALL cell line for preclinical evaluation carrying t(11;19), i.e. a KMT2A::MLLT1 rather than a KMT2A::AFF1 fusion, so it is a newly created resource that is on-topic for KMT2A-rearranged leukaemia and off-target for this beat's index fusion.
Establishment of a t(11;19), KMT2A Rearranged B-ALL Cell Line for Preclinical Evaluation and Novel Therapeutic Discovery · Sharma R, Incoronato A, Zhang C, et al. · PubMed record 2023What this evidence cannot show
Title only; the abstract was not read, the paper was not read, and the author list and DOI were not retrieved for this record. It is listed as a bounded rejection so that a future beat does not have to rediscover it, and nothing about its contents should be quoted from here beyond its title.
Evidence that argues against this (1)
- New cell lines of KMT2A-rearranged B-ALL are being made, which cuts against any framing of this field as one where no new material is generated. The shortage is specific to the index fusion in infants, and it may be a shortage of deposits rather than of derivations. Cruickshank MN et al. 2017
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
PMID 22212479 has PMCID PMC3367544, and on 2026-09-19 neither Europe PMC fullTextXML nor NCBI efetch db=pmc returned a document containing a body element for it, while the same test returned bodies for PMC8082024, PMC11816686, PMC8018117, PMC9883157, PMC8616957, PMC4967476, PMC5063541 and (via efetch after a Europe PMC HTTP 500) PMC6880934.
Full-text retrieval test across eleven beat papers: Europe PMC core record for PMCID, then fullTextXML, then efetch db=pmc, testing for a body element · Oncofusion knowledge graph, Q21 index-fusion-models beat · Europe PMC / NCBI eutils 2026What this evidence cannot show
A single test run on one day. Europe PMC fullTextXML is known from this project's prior experience to return transient HTTP 500s, so a failure on one date is not a permanent unreadability; the correct record is 'not readable by these two routes on 2026-09-19'. The PDF route and publisher site were not tried.
Evidence that argues against this (1)
- This project has twice produced a confident wrong negative from a clean-looking retrieval failure, including an unreadable register in which two of three papers turned out to be open-access full text. One failed fetch of PMC3367544 should be retried before anything is built on its unreadability, and nothing in this beat is. Rokita JL et al. 2019
audit trail for the verdict
- supports →Bounded absence: no published CRISPR or dependency screen in any infant-derived KMT2A::AFF1 model
bounded rejections and unreadables alongside the bounded absence
records PPTC Table S5 as the highest-value unretrieved item