SWI/SNF enumeration for KMT2A-rearranged leukaemia: every subunit searched, every count dated, and what was read and rejected
also called Q47 enumeration · SWI/SNF bounded absences
This is the bookkeeping entry for the SWI/SNF beat: which genes were searched, with which alternative names, what each search returned on which date, which papers looked relevant and were not, and which subunits were deliberately left without an entry.
The same thing, for a scientist
Twenty-four SWI/SNF subunits were enumerated by current symbol, by BAF-nomenclature alias and at complex level, on PubMed via E-utilities on **2026-09-20**. Every zero was checked against separately-counted operands and a positive control through the identical code path, per rule 67 [E1]. Every subunit belonging to a named complex was also searched at complex level, per rule 64 [E2]. The headline is that the literature on SWI/SNF in this disease is **very small and heavily front-loaded on three subunits**: `SMARCD2 AND (MLL OR KMT2A)` = 1, `DPF2 AND (MLL OR KMT2A)` = 1, `BICRA`/`GLTSCR1 AND (MLL OR KMT2A)` = 0, `SMARCE1 AND leukemia` = 0; while the apparently larger counts are mostly noise - three of the fourteen `SMARCA4 AND (MLL OR KMT2A)` records were fetched in full and all three were off-beat [E1][E3]. Ten entries were written; **four subunits were deliberately not given entries** and are recorded here instead [E4].
Counts, with aliases, as run on 2026-09-20
| query | records | note |
|---|---|---|
SMARCA4 AND (MLL OR KMT2A) | 14 | 3 fetched, all off-beat; on-beat primaries are 26571505 and 40672274 |
BRG1 AND (MLL OR KMT2A) | 11 | alias run |
SMARCA2 AND (MLL OR KMT2A) | 6 | no primary SMARCA2 study |
BRM AND (MLL OR KMT2A) | 5 | short word-like alias, not title-checked |
SMARCB1 AND (MLL OR KMT2A) | 11 | title-checked to 0 on-beat |
SNF5 / INI1 / BAF47 AND (MLL OR KMT2A) | 4 / 4 / 4 | aliases |
ARID1A AND (MLL OR KMT2A) | 35 | dominated by co-mention cohorts |
BAF250A AND leukemia | 16 | alias |
ARID2 AND (MLL OR KMT2A) | 9 | 2 read, both on-beat |
BAF200 AND leukemia | 2 | alias |
SMARCD2 AND (MLL OR KMT2A) | 1 | the anchor only |
BAF60B AND leukemia | 5 | alias recovered PMID 29483235, invisible to the current symbol |
SMARCD1 AND (MLL OR KMT2A) | 1 | fetched: off-beat gene-list paper |
BAF60A AND (MLL OR KMT2A) | 0 | alias |
DPF2 AND (MLL OR KMT2A) | 1 | the anchor only |
BAF45D AND leukemia | 1 | the anchor only |
SMARCE1 AND (MLL OR KMT2A) | 1 | fetched: off-beat |
SMARCE1 AND leukemia | 0 | operands checked: SMARCE1 = 189, SMARCE1 AND cancer = 101 |
BAF57 AND leukemia | 2 | alias recovers 2 where the current symbol returns 0 |
ACTL6A AND (MLL OR KMT2A) | 1 | not fetched |
ACTL6A AND leukemia | 3 | not fetched |
SMARCC1 AND (MLL OR KMT2A) | 1 | fetched: glioblastoma ASH2L screen, off-beat |
SMARCC2 AND (MLL OR KMT2A) | 1 | not fetched |
BICRA AND (MLL OR KMT2A) | 0 | operands: BICRA = 28, BICRA AND cancer = 12 |
GLTSCR1 AND (MLL OR KMT2A) | 0 | operands checked |
SS18 AND (MLL OR KMT2A) | 3 | not fetched |
SWI/SNF AND (MLL OR KMT2A) | 27 | complex form, rule 64 |
BAF complex AND (MLL OR KMT2A) | 2 | complex form |
ncBAF AND leukemia | 15 | complex form, not title-checked |
PBAF AND leukemia | 14 | complex form, not title-checked |
BAF AND (MLL-AF4 OR MLL-AF9 OR KMT2A-AFF1) AND immunoprecipitation | 0 | operands: MLL-AF9 AND immunoprecipitation = 28, SWI/SNF AND immunoprecipitation = 317 |
Positive controls through the identical code path: leukemia = 405,093 and menin AND KMT2A = 271. The instrument returns non-zero 1.
Bounded rejections - read and judged off-beat
Papers a reader would reasonably expect to be cited here and which are not, with the reason. All were fetched and their abstracts read on 2026-09-20.
- PMID 32341579, Locoregionally administered B7-H3-targeted CAR T cells for treatment of atypical teratoid/rhabdoid tumors (Nat Med 2020). Returned by both
SMARCA4 AND (MLL OR KMT2A)andSMARCB1 AND (MLL OR KMT2A). It is an ATRT CAR-T paper; SMARCA4 appears because residual BRG1 activity sustains B7-H3 expression. Would change my mind: if a KMT2A-rearranged cohort were included. - PMID 33659785, Longitudinal genetic studies of cognitive characteristics (Vavilovskii Zhurnal 2020). Returned by SMARCA4, SMARCB1, SMARCE1, SMARCA2 and ARID1A queries. A review of cognitive genetics that lists chromatin genes. Pure co-mention noise.
- PMID 24522270, Chromatin-regulating proteins as targets for cancer therapy (J Radiat Res 2014). A radiotherapy-oriented review; mentions DOT1L in MLL-rearranged leukaemia and BRM inhibition in BRG1-deficient tumours but contains no SWI/SNF-in-MLL result. Used only as background, not cited.
- PMID 37974198, Epigenetic-focused CRISPR/Cas9 screen identifies ASH2L as a regulator of glioblastoma cell survival (Cell Commun Signal 2023). The sole hit of
SMARCC1 AND (MLL OR KMT2A). SMARCC1 appears as a gene downregulated after ASH2L depletion in glioblastoma. Off-beat. - PMID 38114583, De novo variants underlying monogenic syndromes with intellectual disability (Eur J Hum Genet 2024). The sole hit of
SMARCD1 AND (MLL OR KMT2A). Both genes appear in a 46-gene list. Off-beat. - PMID 41155355, ATRT in an MEN1 patient case report. Returned by
SMARCA4 AND menin. A clinical case report; the MEN1 in it is the syndrome, not the menin-KMT2A axis. Off-beat, and a nice example of a symbol collision between a gene and a syndrome name. - PMID 29958106, CBFbeta-SMMHC Inhibition Triggers Apoptosis by Disrupting MYC Chromatin Dynamics in AML (Cell 2018). On-topic for SWI/SNF and MYC enhancers - RUNX1 replaces BRG1 with RING1B at three MYC distal enhancers - but it is inv(16) leukaemia, not KMT2A. Not cited here, and it is the best candidate for a follow-up beat, because it is the only paper found that manipulates BRG1 occupancy at MYC enhancers pharmacologically in a fusion-driven leukaemia.
Subunits deliberately not given an entry, with reason: SMARCE1 (SMARCE1 AND leukemia = 0, alias BAF57 = 2, no KMT2A-relevant evidence, DepMap rank 7,543 with difference -0.012); ACTL6A (DepMap rank 17,410, difference +0.145, pan-essentiality -1.57 - floor-compressed and anti-selective); SMARCC1/SMARCC2 (one off-beat record each, no selectivity); PBRM1, PHF10, BRD7 (PBAF, all flat, no KMT2A literature). Writing any of these as a roster entry would have padded the beat without adding evidence.
| script | manual PubMed E-utilities enumeration; helper at scratchpad q47_pubmed.py, not committed |
|---|---|
| status | run |
| date | 2026-09-20 |
| subunits enumerated | 24 |
inputs
[ "NCBI E-utilities esearch", "UniProt REST gene_exact queries", "analyses/depmap-kmt2a-selective-scan-full/results.json", "analyses/swi-snf-lineage-split/results.json" ]
entries written
[ "gene-smarca4", "gene-smarca2", "gene-smarcd2", "gene-smarcd1", "gene-dpf2", "gene-smarcb1", "gene-arid1a", "gene-arid2", "gene-bicra", "gene-ss18" ]
subunits deliberately not written
[
{
"gene": "SMARCE1",
"reason": "`SMARCE1 AND leukemia` = 0 (operands checked); alias BAF57 returns 2, unread; DepMap rank 7,543, difference -0.012",
"what_would_change_it": "any leukaemia loss-of-function result, or a title-check of the two BAF57 records"
},
{
"gene": "ACTL6A",
"reason": "DepMap rank 17,410, difference +0.145, pan-essentiality -1.57; floor-compressed and anti-selective",
"what_would_change_it": "a KMT2A-rearranged-specific ACTL6A result, or evidence that ACTL6B substitutes in this lineage"
},
{
"gene": "SMARCC1/SMARCC2",
"reason": "one off-beat record each; no selectivity (ranks 8,543 and 2,474 with differences -0.005 and -0.066)",
"what_would_change_it": "the Chatterjee SMARCC1-nucleated residual complex being shown in KMT2A-rearranged blasts specifically"
},
{
"gene": "PBRM1 / PHF10 / BRD7",
"reason": "PBAF-specific, all flat in both scans, no KMT2A literature; PBAF is covered by gene-arid2",
"what_would_change_it": "a PBAF-resolved chromatin study in this disease"
}
]bounded rejections — the numbers behind the chart
[
{
"pmid": "32341579",
"title_gist": "B7-H3 CAR-T in ATRT",
"reason": "SMARCA4 appears as residual BRG1 activity sustaining B7-H3 in a rhabdoid tumour; no leukaemia",
"what_would_change_it": "inclusion of a KMT2A-rearranged cohort",
"title": "Locoregionally administered B7-H3-targeted CAR T cells for treatment of atypical teratoid/rhabdoid tumors.",
"authors": "Theruvath J, Sotillo E, Mount CW, et al.",
"venue": "Nat Med",
"year": 2020,
"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."
},
{
"pmid": "33659785",
"title_gist": "longitudinal genetics of cognitive characteristics",
"reason": "review listing chromatin genes; pure co-mention",
"what_would_change_it": "nothing short of a different paper",
"title": "Longitudinal genetic studies of cognitive characteristics.",
"authors": "Mustafin RN, Kazantseva AV, Enikeeva RF, et al.",
"venue": "Vavilovskii Zhurnal Genet Selektsii",
"year": 2020,
"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."
},
{
"pmid": "24522270",
"title_gist": "chromatin-regulating proteins as cancer targets",
"reason": "radiotherapy-oriented review, no SWI/SNF-in-MLL result",
"what_would_change_it": "n/a; used as background only",
"title": "Chromatin-regulating proteins as targets for cancer therapy.",
"authors": "Oike T, Ogiwara H, Amornwichet N, et al.",
"venue": "J Radiat Res",
"year": 2014,
"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."
},
{
"pmid": "37974198",
"title_gist": "ASH2L in glioblastoma",
"reason": "SMARCC1 appears as a downstream downregulated gene in glioblastoma",
"what_would_change_it": "a leukaemia replication",
"title": "Epigenetic-focused CRISPR/Cas9 screen identifies (absent, small, or homeotic)2-like protein (ASH2L) as a regulator of glioblastoma cell survival.",
"authors": "Ozyerli-Goknar E, Kala EY, Aksu AC, et al.",
"venue": "Cell Commun Signal",
"year": 2023,
"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."
},
{
"pmid": "38114583",
"title_gist": "de novo variants in intellectual disability",
"reason": "SMARCD1 and KMT2A in a 46-gene list",
"what_would_change_it": "nothing",
"title": "De novo variants underlying monogenic syndromes with intellectual disability in a neurodevelopmental cohort from India.",
"authors": "Pande S, Majethia P, Nair K, et al.",
"venue": "Eur J Hum Genet",
"year": 2024,
"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."
},
{
"pmid": "41155355",
"title_gist": "ATRT in an MEN1 patient",
"reason": "MEN1 the syndrome, not the menin-KMT2A axis",
"what_would_change_it": "nothing",
"title": "Unusual Dual Brain Tumor Morphologies in an MEN1 Patient: A Case Report of Diagnostic Challenges and Methylation Insights.",
"authors": "Patel V, Aboud O, Barakat A",
"venue": "Int J Mol Sci",
"year": 2025,
"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."
},
{
"pmid": "29958106",
"title_gist": "CBFb-SMMHC inhibition disrupts MYC chromatin dynamics",
"reason": "inv(16) AML, not KMT2A - but it replaces BRG1 with RING1B at three MYC distal enhancers and is the strongest adjacent result found",
"what_would_change_it": "nothing; it is recommended as the seed of a follow-up beat rather than rejected",
"title": "CBFβ-SMMHC Inhibition Triggers Apoptosis by Disrupting MYC Chromatin Dynamics in Acute Myeloid Leukemia.",
"authors": "Pulikkan JA, Hegde M, Ahmad HM, et al.",
"venue": "Cell",
"year": 2018,
"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."
}
]- A beat that leaves four subunit groups unwritten is making a prediction that they do not matter; if a canonical-BAF dependency exists that single-gene CRISPR cannot see, those are exactly the entries that should have been written.
- Every count here is dated 2026-09-20 and will drift; a reader in a year should re-run rather than cite.
- Does the inv(16) MYC-enhancer result (PMID 29958106) generalise - is BRG1-to-RING1B replacement at MYC enhancers a shared consequence of shutting down any fusion oncoprotein, and does menin inhibition do it in KMT2A-rearranged cells?
- What do the two `BAF57 AND leukemia` records say? SMARCE1 was left unwritten on a count its own alias contradicts.
- Would a full-text search rather than an abstract search change these counts? Every number here is an abstract-level bound.
- E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
Twenty-four SWI/SNF subunits were enumerated on PubMed by current symbol and BAF alias on 2026-09-20; the counts are as tabulated, and every zero was checked against separately-counted operands and a positive control through the identical code path.
PubMed E-utilities esearch record counts: full enumeration of 24 SWI/SNF subunits by current symbol and BAF-nomenclature alias against `(MLL OR KMT2A)` and against `leukemia`, plus complex forms `SWI/SNF`, `BAF complex`, `ncBAF`, `PBAF`; operand and positive controls `leukemia` = 405,093, `menin AND KMT2A` = 271, `SWI/SNF AND immunoprecipitation` = 317, `MLL-AF9 AND immunoprecipitation` = 28; run 2026-09-20 via NCBI E-utilities esearch · NCBI E-utilities, run for this beat on 2026-09-20 · PubMed 2026What this evidence cannot show
Bounded absence. Bounded to PubMed title/abstract/all-fields on 2026-09-20. Counts are raw esearch counts; only eleven records across the whole beat were fetched and read in full, so most on-beat/off-beat judgements are from titles. Short word-like aliases (BRM, REQ, SYT, SEPT-style tokens) were deliberately not searched because this project has measured them inflating counts up to 57-fold.
Evidence that argues against this (1)
- A raw count is an upper bound contaminated by co-mention and a lower bound for complex-framed work; both errors are present here in the same table, which is why the fetched examples matter more than the numbers. Shi J et al. 2013
- E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
Complex-level forms were run for every subunit per rule 64: `SWI/SNF AND (MLL OR KMT2A)` = 27, `BAF complex AND (MLL OR KMT2A)` = 2, `ncBAF AND leukemia` = 15, `PBAF AND leukemia` = 14.
PubMed E-utilities esearch record counts: complex-level forms run 2026-09-20: `SWI/SNF AND (MLL OR KMT2A)` = 27, `BAF complex AND (MLL OR KMT2A)` = 2, `ncBAF AND leukemia` = 15, `PBAF AND leukemia` = 14 · NCBI E-utilities, run for this beat on 2026-09-20 · PubMed 2026What this evidence cannot show
The `ncBAF` and `PBAF` result sets were NOT individually title-checked, so a complex-level result in KMT2A-rearranged disease naming no subunit could still have been missed. 'PBAF' and 'BAF' are short tokens subject to ordinary-word contamination, so both counts are upper bounds.
Evidence that argues against this (1)
- The complex form found nothing the symbol queries missed in this beat, but the same check in a previous beat overturned a wrong absence claim, so a null here is weak evidence that the practice is unnecessary. Michel BC et al. 2018
- E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
The apparently larger symbol counts are heavily contaminated: of the SMARCA4, SMARCB1, SMARCC1 and SMARCD1 conjunction hits that were fetched and read, every one was off-beat - an ATRT CAR-T study, a cognitive-genetics review, a glioblastoma ASH2L screen and a neurodevelopmental gene list.
SMARCB1 Deficiency Integrates Epigenetic Signals to Oncogenic Gene Expression Program Maintenance in Human Acute Myeloid Leukemia. · Chatterjee SS, Biswas M, Boila LD, Banerjee D, et al. · Mol Cancer Res 2018What this evidence cannot show
Six records were fetched for this purpose; the claim is about those six and is generalised to the remainder only as a caution. The citation attached is the one substantive SWI/SNF-in-AML paper recovered from the same searches, which is the contrast being drawn.
Evidence that argues against this (1)
- Judging the unfetched remainder by the fetched sample is exactly the inference this project has been burned by; the honest statement is that the on-beat fraction is unknown and demonstrably below 1. Cruickshank VA et al. 2015
- E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.dataset
Four subunit groups were deliberately left without entries because no KMT2A-relevant evidence exists for them: SMARCE1, ACTL6A, SMARCC1/SMARCC2 and the PBAF-specific PBRM1/PHF10/BRD7.
Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap Chronos via Open Targets) · Oncofusion knowledge graph, analyses/depmap-kmt2a-selective-scan-full/run.py · DepMap Chronos gene effect via Open Targets GraphQL 2026What this evidence cannot show
The decision rests on a zero or near-zero literature plus a flat DepMap difference; ACTL6A in particular is at -1.57 pan-essentiality, where a flat difference is uninterpretable rather than informative, so 'no evidence of selectivity' is the correct reading and 'not required' would be wrong.
Evidence that argues against this (1)
- SMARCE1 returns 0 under its current symbol and 2 under BAF57, so at least one of these decisions was made on a count that its own alias contradicts; the two BAF57 records were not read. NCBI E-utilities et al. 2026
the dependency half of the same beat's evidence base
SMARCB1's bounded absence: 11 records, title-checked to 0 on-beat
BICRA's zero, with operands and positive control
that entry recorded PMID 26571505 as an unfilled gap; this beat fills it
SWI/SNF is the sixth instance of the named-but-unwritten pattern that entry describes