AnalysismoderateQ5-interactome4

Interactome round 4: what was counted, what was written, what was rejected, and the two alias traps this beat fell into and caught

also called interactome4 enumeration · round 4 enumeration ledger

Before writing anything, this round counted how many published papers exist for each candidate protein in KMT2A leukaemia. Most of the proteins the brief asked about turned out to have almost no literature at all, and two of the searches produced wildly wrong counts because of two-letter gene nicknames.

The same thing, for a scientist

🟠 **ARM-COMPOSITION CAVEAT ADDED 2026-09-24 (cycle 140).** This entry's computation rests on `analyses/depmap-kmt2a-selective-scan-full`, and that scan **fixes both comparison arms from the first gene that returns screen data** (`run.py:170`, `if rear is None:`; line 174 prints *"Groups fixed from first gene with data"*). Because DepMap screens each gene in a gene-specific set of lines, **5 of 16 eligible KMT2A-rearranged lines are excluded from every gene comparison, including genes where they were screened** - 16 resolve under the scan's own query, 11 were used. **This does not show any conclusion here is wrong.** Dropping 5 of 16 costs power, and it biases a ranking only if the dropped lines differ systematically from the kept ones, **which nobody has tested.** The named one-line fix is to derive the arms from each gene's own screen list. Until the scan is re-run, selectivity ranks taken from it carry an unquantified arm-composition caveat. Found by `analyses/depmap-scan-line-identities` on 2026-09-21 and unreported until now; see `ana-the-depmap-selectivity-scan-fixes-its-arms-from-the-first-gene-and-drops-five-of-sixteen-lines-from-every-comparison`. --- This is the enumeration ledger for interactome round 4. **41 PubMed record-count queries** were run through the NCBI E-utilities esearch endpoint on 2026-09-19, together with a genome-wide read of `analyses/depmap-kmt2a-selective-scan-full` for every candidate [E1][E2]. The headline result is that **the beat's dispatch premise was half wrong**: the brief asked for DSIF and NELF on the grounds that 'pausing is half the story of an elongation hijack and it is absent'. It is absent, but from the **field**, not from this graph — `(MLL OR KMT2A) AND (SUPT5H OR SPT5 OR SPT5H OR DSIF)` returns **1**, `SUPT4H1/SPT4` returns **0**, and `(MLL OR KMT2A) AND NELF` returns **1**, a CDK9 review [E1]. Worse for the premise, two published results argue pause release is **not** the rate-limiting step in MLL fusion-dependent transcription [E3][E4]. **20 entries were written**; **14 candidates were rejected or deferred**, each with its count and reason below. Two enumeration queries had to be discarded for alias inflation, which is a methodological finding in its own right [E5].

Go deeper

What was written (20 entries)

Entry	Block	Why it earned an entry
`gene-pcgf1`	PRC1.1	menin- and DOT1L-inhibitor resistance gene; 4 records
`gene-rnf2`	PRC1.1	PRC1 catalytic core; Ring1a/b required for MLL-AF9 leukaemogenesis; 8 records
`gene-bcorl1`	PRC1.1	in the only PRC1.1 core structure; 7 records, none functional
`gene-skp1`	PRC1.1	structural subunit; pan-essential dead end; 1 record
`protein-prc1-1-complex`	PRC1.1	complex-level synthesis; the subunit-by-subunit dependency pattern
`gene-supt5h`	pausing	counted absence + the 7OKX DSIF-ELL2-EAF1 structure
`gene-supt4h1`	pausing	0 records, rank 138/17,728 - an unexplained lead
`protein-nelf-complex`	pausing	counted absence + per-subunit dependency table
`gene-nelfb`	pausing	most extreme anti-dependency in the beat
`gene-nelfe`	pausing	RRM subunit; the 'RD' alias trap
`hyp-pause-axis-asymmetry-kmt2a-r`	pausing	six-gene asymmetry, falsifiable, prior 0.2
`exp-pro-seq-pausing-index-kmt2a-aff1`	pausing	the one experiment that settles it
`gene-eaf1`	ELL/EAF	Cajal body localisation lost on MLL-ELL - the only subnuclear result in the beat
`gene-eaf2`	ELL/EAF	mapped ELL-binding tumour-suppressor module
`gene-ice1`	LEC	explains ELL's pan-essentiality away from the fusion; 0 records
`gene-ice2`	LEC	completes LEC; 0 records
`gene-ep300`	tier 5	rank 108; paralogue split against CREBBP at rank 16,950
`gene-kdm1a`	tier 5	acts at MLL-AF9-bound loci, spares normal HSPC
`gene-myb`	tier 5	rank 17; partial transient suppression eradicates AML without stopping myelopoiesis
`gene-brd9`	tier 5	rank 39 with positive pan-essentiality; on-target swap-allele proof

What was rejected or deferred, with counts and reasons

Candidate	Records (query)	Decision	Reason
NELFA	0 `(MLL OR KMT2A) AND (NELFA OR WHSC2 OR NELF-A)`	recorded in `protein-nelf-complex`, no separate entry	every field would read 'not established'; its extreme DepMap rank (17,650) is floor-compressed and uninterpretable alone
NELFCD	0 `(MLL OR KMT2A) AND (NELFCD OR TH1L OR NELF-C OR NELF-D)`	recorded in `protein-nelf-complex`, no separate entry	no distinguishing evidence; dependency unremarkable (36th percentile)
RYBP / YAF2	1 `(MLL OR KMT2A) AND (RYBP OR YAF2 OR DEDAF)`	deferred	genuine variant-PRC1 subunit and the determinant of RING1B stimulation, but absent from the DepMap scan and with one record; would have been an entry of two facts
USP7	2 `(MLL OR KMT2A) AND USP7 AND leukemia`	deferred	PRC1.1-associated in some preparations; pan-essential-adjacent (-0.847) with no selectivity (rank 4,242); a drug-target story that belongs to a therapeutics beat
CXXC1 / CFP1	8	deferred	rank 414 (-0.161) is a real signal and the CpG-reader parallel to the fusion's retained CXXC domain is interesting; deferred for want of a primary paper tying it to a KMT2A fusion
SETD1B	17	deferred	rank 1,311 (-0.093); the strongest primary paper found is B-cell lymphoma (venetoclax resistance via BCL2-family proapoptotic genes), not KMT2A leukaemia
CDK12	2 (with CDK13)	rejected	rank 17,136, difference +0.115 - anti-dependency
CDK13	2 (with CDK12)	deferred	rank 171 (-0.238) is a strong signal and the CDK12/CDK13 paralogue split mirrors EP300/CREBBP; no KMT2A-specific literature at all
ZMYND11 / BS69	4	rejected	rank 4,016, difference -0.043; no selectivity
CHD4 / NuRD	12	rejected	rank 13,245, difference +0.034 on a pan-essentiality of -1.117 - pan-essential and anti-selective
MTA2	0 `(MLL OR KMT2A) AND (MTA2 OR MTA1 OR MTA3)`	rejected	zero literature and rank 1,037 (-0.105), inside noise
HDAC1 / HDAC2	25	rejected	both anti-dependencies: HDAC1 +0.103 (rank 16,942), HDAC2 +0.115 (rank 17,134)
SIN3A	5	rejected	rank 17,096, difference +0.111 on pan-essentiality -0.883
SMARCA4 / BRG1	21	rejected	rank 16,333, difference +0.079 - anti-dependency; ncBAF is represented in this beat by BRD9 alone
ARID1A	36	rejected	rank 12,593, difference +0.028; also a canonical-BAF rather than ncBAF subunit
SPI1 / PU.1	7 `(MLL OR KMT2A) AND SPI1`	deferred with regret	rank 43 (-0.385) on a pan-essentiality of -0.091 - one of the best-shaped dependencies in the scan, and the Leucegene transcriptomic study reports a subset of MLL-fusion AML carrying SPI1 mutations with inactivation of its transcriptional network. Deferred only for want of writing time; this is the first candidate a round 5 should pick up
CEBPA	143 `(MLL OR KMT2A) AND CEBPA`	deferred	rank 25 (-0.438) on pan-essentiality -0.169 - excellent shape, and a large literature that could not be read in the time available. Second candidate for round 5

Pattern worth naming. Of the fourteen rejections, seven were rejected because the gene is an anti-dependency in KMT2A-rearranged lines (CDK12, CHD4, HDAC1, HDAC2, SIN3A, SMARCA4, ARID1A), and every one of those is a chromatin repressor or remodeller. Meanwhile the three deferrals with the best dependency shape (SPI1, CEBPA, MYB — all written or flagged) are myeloid transcription factors. The beat went looking for chromatin complexes and the dependency data kept pointing at transcription factors instead.

The two alias traps, and the rule they imply

The standing brief instructs agents to search retired symbols, because for the septin fusion partners the retired symbol returned twenty records where the current symbol returned one. That rule is about recall. This beat hit the opposite failure twice, both times about precision 5:

Query	Records	Corrected query	Records	Inflation
`(MLL OR KMT2A) AND (NELFE OR RDBP OR RD OR "NELF-E")`	115	drop the alias `RD`	2	57x
`(MLL OR KMT2A) AND (EAF2 OR U19 OR TRAITS)`	75	drop `U19` and `TRAITS`	5	15x

Both would have supported a confident, wrong statement that the protein is well studied in this disease. The existing rule in the brief names the SEPT* wildcard trap (matching 'September'); these are the same hazard arriving through an explicit alias rather than a wildcard, which the current wording does not cover.

Proposed addition to the standing brief: an alias that is one or two characters, or that is an ordinary English word or abbreviation (RD, U19, TRAITS, RD, ICE1, ICE2, SPT4), must have its records title-checked before its count is used. Alias lists need a precision pass as well as a recall pass.

A third, different trap also appeared: (ICE1 OR ICE2) AND "little elongation complex" returns 7, of which at least three are incidental mentions in a heparin-induced-thrombocytopenia GWAS, a cleft-palate parent-of-origin analysis and a skeletal-muscle extracellular-vesicle proteomics study 6. Small counts must be title-checked in both directions — for inflation and for irrelevance.

The premise in the dispatch brief that the literature contradicts

The brief listed DSIF and NELF as structural hole number two, reasoning that 'pausing is half the story of an elongation hijack'. The enumeration says the field does not agree, and two primary results say so explicitly:

Okuda and Yokoyama 2016 name the hypothesis — 'it is hypothesized that relieving the promoter proximal pausing of RNA polymerase II is the rate-limiting step of MLL fusion-dependent transcription' — and report that TBP loading through SL1 is the rate-limiting activity, not elongation 3.
Xiang et al. 2026 report that the AFF4-ELL2-EAF1-containing ECAS complex, whose perturbation impairs MLL-rearranged leukaemia stem cell function, acts without altering pause release 4.

The pausing entries were still written, for three reasons: the counted absence is itself worth recording; the graph's P-TEFb axis had no substrate entry until gene-supt5h supplied SPT5's CTR repeats; and the dependency data produced an unexpected asymmetry that is now a falsifiable hypothesis with one experiment attached. But the graph should not carry 'the fusion hijacks pause release' as settled, and after this cycle it does not.

Facts and figures

script	no script; PubMed E-utilities esearch queries and a scripted read of analyses/depmap-kmt2a-selective-scan-full/results.json
status	run
n queries	41
n entries written	20
n candidates rejected or deferred	16
result summary	The dispatch brief's second structural hole (DSIF and NELF) is a gap in the field rather than in the graph: 1 record for DSIF, 0 for SPT4, 1 for NELF. Two published results place the rate-limiting step of MLL fusion-dependent transcription away from pause release. Seven of the rejected candidates were rejected because they are ANTI-dependencies in KMT2A-rearranged lines, and all seven are chromatin repressors or remodellers, while the best-shaped dependencies found are myeloid transcription factors (MYB rank 17, CEBPA 25, SPI1 43). Two alias-inflated queries (57-fold and 15-fold) had to be discarded, which implies a precision rule to sit alongside the brief's existing recall rule.
arm composition caveat cycle 140	source scan fixes both arms from the first gene with data (run.py:170); 5 of 16 eligible KMT2A-rearranged lines are excluded from every comparison. Effect on this entry's conclusion untested.

rejections

inputs

[
  "NCBI PubMed E-utilities esearch, 41 queries, 2026-09-19",
  "analyses/depmap-kmt2a-selective-scan-full/results.json (generated 2026-09-19T10:12:06Z; 17,728 genes, 1,208 lines, 11 KMT2A-rearranged vs 43 other leukaemia)",
  "UniProt REST gene_exact lookups with primary-gene-name read-back for 20 accessions",
  "RCSB data API for PDB 5JH5, 7OKX and 7OKY"
]

rejections — the numbers behind the chart

[
  {
    "candidate": "NELFA",
    "records": 0,
    "depmap_rank": 17650,
    "decision": "folded into protein-nelf-complex",
    "reason": "no distinguishing evidence; extreme rank is floor-compressed"
  },
  {
    "candidate": "NELFCD",
    "records": 0,
    "depmap_rank": 6425,
    "decision": "folded into protein-nelf-complex",
    "reason": "no distinguishing evidence; dependency unremarkable"
  },
  {
    "candidate": "RYBP/YAF2",
    "records": 1,
    "depmap_rank": null,
    "decision": "deferred",
    "reason": "absent from the DepMap scan; one record; would be an entry of two facts"
  },
  {
    "candidate": "USP7",
    "records": 2,
    "depmap_rank": 4242,
    "decision": "deferred",
    "reason": "no selectivity; belongs to a therapeutics beat"
  },
  {
    "candidate": "CXXC1/CFP1",
    "records": 8,
    "depmap_rank": 414,
    "decision": "deferred",
    "reason": "real dependency signal but no primary paper tying it to a KMT2A fusion"
  },
  {
    "candidate": "SETD1B",
    "records": 17,
    "depmap_rank": 1311,
    "decision": "deferred",
    "reason": "strongest primary paper is B-cell lymphoma venetoclax resistance, not KMT2A leukaemia"
  },
  {
    "candidate": "CDK12",
    "records": 2,
    "depmap_rank": 17136,
    "decision": "rejected",
    "reason": "anti-dependency, +0.115"
  },
  {
    "candidate": "CDK13",
    "records": 2,
    "depmap_rank": 171,
    "decision": "deferred",
    "reason": "strong signal (-0.238) with zero KMT2A literature; a paralogue split mirroring EP300/CREBBP"
  },
  {
    "candidate": "ZMYND11/BS69",
    "records": 4,
    "depmap_rank": 4016,
    "decision": "rejected",
    "reason": "no selectivity, -0.043"
  },
  {
    "candidate": "CHD4/NuRD",
    "records": 12,
    "depmap_rank": 13245,
    "decision": "rejected",
    "reason": "pan-essential (-1.117) and anti-selective (+0.034)"
  },
  {
    "candidate": "MTA2",
    "records": 0,
    "depmap_rank": 1037,
    "decision": "rejected",
    "reason": "zero literature; -0.105 inside noise"
  },
  {
    "candidate": "HDAC1",
    "records": 25,
    "depmap_rank": 16942,
    "decision": "rejected",
    "reason": "anti-dependency, +0.103"
  },
  {
    "candidate": "HDAC2",
    "records": 25,
    "depmap_rank": 17134,
    "decision": "rejected",
    "reason": "anti-dependency, +0.115"
  },
  {
    "candidate": "SIN3A",
    "records": 5,
    "depmap_rank": 17096,
    "decision": "rejected",
    "reason": "anti-dependency, +0.111"
  },
  {
    "candidate": "SMARCA4/BRG1",
    "records": 21,
    "depmap_rank": 16333,
    "decision": "rejected",
    "reason": "anti-dependency, +0.079"
  },
  {
    "candidate": "ARID1A",
    "records": 36,
    "depmap_rank": 12593,
    "decision": "rejected",
    "reason": "no selectivity (+0.028); canonical BAF rather than ncBAF"
  },
  {
    "candidate": "SPI1/PU.1",
    "records": 7,
    "depmap_rank": 43,
    "decision": "deferred with regret",
    "reason": "excellent shape (-0.385 on pan-essentiality -0.091) plus SPI1 mutations reported in a subset of MLL-fusion AML; deferred only for time. FIRST candidate for round 5"
  },
  {
    "candidate": "CEBPA",
    "records": 143,
    "depmap_rank": 25,
    "decision": "deferred",
    "reason": "excellent shape (-0.438 on pan-essentiality -0.169) and a 143-record literature that could not be read in time. SECOND candidate for round 5"
  }
]

method findings

[
  "Alias precision is a distinct failure mode from alias recall and the standing brief covers only the latter. RD inflated a count 57-fold; U19 and TRAITS inflated another 15-fold.",
  "Small counts must be title-checked for irrelevance as well as inflation: 3 of 7 ICE1/ICE2 records were incidental mentions.",
  "A gene-symbol query cannot find a protein that appears only in a PDB entity list. PDB 7OKX places SPT5 with ELL2 and EAF1 and was invisible to every DSIF query in this beat."
]

left for round 5

[
  "SPI1",
  "CEBPA",
  "CDK13",
  "CXXC1",
  "RYBP",
  "BICRA/BICRAL (ncBAF-defining subunits, never counted)",
  "permutation and lineage-conditioning of every dependency claim made this cycle"
]

How this could bite us back

This ledger ranks candidates largely on one dependency statistic from one dataset over eleven cell lines. If that statistic is confounded, the whole rejection list is wrong in the same direction at once.
Rejecting a gene because it is an anti-dependency assumes the relevant phenotype is proliferation. For a resistance gene, a self-renewal gene or a differentiation gene, the assumption fails - and PCGF1, which this beat wrote up precisely because its loss confers drug resistance, is the proof that it fails.
Two deferrals (SPI1, CEBPA) were made for writing time rather than for evidence, which is an honest but poor reason and leaves two of the best-shaped dependencies in the scan undocumented.

Still open

Should the standing brief carry an alias-precision rule alongside the alias-recall rule? This beat hit the precision failure twice in one cycle.
Why are seven of the rejected chromatin repressors and remodellers ANTI-dependencies in KMT2A-rearranged lines, while the myeloid transcription factors are the strongest dependencies? That is a pattern across fourteen genes and it has not been analysed.
Would permutation, lineage conditioning and line normalisation survive for MYB, BRD9, SPI1, CEBPA, EP300 and SUPT4H1? None was run this cycle and all six claims depend on it.
How many other interactors are hiding in PDB entity lists rather than in gene-symbol literature? A systematic RCSB search over KMT2A-relevant complexes has never been done in this project.

Every source behind this entry (6)

E1strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
Forty-one PubMed record-count queries were run through the NCBI E-utilities esearch endpoint on 2026-09-19 for this beat. Key counts against '(MLL OR KMT2A)': RNF2/RING1B/RING1A/RING1 8; PCGF1/NSPC1/RNF68 4; BCORL1/CXXC10 7; SKP1/SKP1A/OCP2/TCEB1L 1; RYBP/YAF2/DEDAF 1; SUPT5H/SPT5/SPT5H/DSIF 1; SUPT4H1/SPT4/SUPT4H 0; NELF 1; NELFA/WHSC2 0; NELFB/COBRA1 2; NELFCD/TH1L 0; NELFE/RDBP 2; EAF1 10; EAF2 5; ICE1/KIAA0947 0; ICE2/NARG2/C16orf80 0; EP300/p300 with leukemia 66; KDM1A/LSD1/AOF2/BHC110 41; SETD1B/KMT2G/KIAA1076 17; CXXC1/CFP1/CGBP 8; CDK12/CDK13/CRKRS/CDC2L5 2; ZMYND11/BS69 4; CHD4/Mi-2beta/NuRD 12; SIN3A 5; HDAC1/HDAC2 25; MTA2/MTA1/MTA3 0; SMARCA4/BRG1/BAF190 21; ARID1A/BAF250A/SMARCF1 36; BRD9 3; SPI1 7; CEBPA 143; MYB/c-Myb 65.
PubMed E-utilities esearch record counts for 41 candidate interactor queries against KMT2A, MLL and MLL-fusion terms, run for interactome round 4 · NCBI PubMed, queried 2026-09-19 2026
What this evidence cannot show
Counts are from the esearch `count` field and were not individually title-checked except where stated. PubMed term mapping expands many gene symbols to supplementary-concept records, so a query is not a literal string match and two queries that look equivalent can differ. Every count here is a **floor** for recall and an **upper bound** for relevance.
Evidence that argues against this (1)
- This beat demonstrated on itself that the counts are fragile in both directions: two queries were inflated 57-fold and 15-fold by generic aliases, and one small count was three-sevenths incidental mentions. Any conclusion drawn from a single count in this table without reading the records should be treated as provisional. NCBI PubMed, queried 2026-09-19 2026
E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.database
Every candidate in this beat was looked up in the project's genome-wide KMT2A-rearranged selectivity scan of 17,728 genes across 1,208 cell lines (11 KMT2A-rearranged versus 43 other leukaemia lines). Selected results by rank: MEN1 1 (-0.8656), MYB 17 (-0.4746), CEBPA 25 (-0.4377), BRD9 39 (-0.3896), SPI1 43 (-0.3850), EP300 108 (-0.2837), SUPT4H1 138 (-0.2642), CDK13 171 (-0.2383), KDM1A 331 (-0.1762), CXXC1 414 (-0.1606), SUPT5H 767 (-0.1215), MTA2 1,037 (-0.1047), SETD1B 1,311 (-0.0931), ICE2 1,775 (-0.0795), ZMYND11 4,016 (-0.0434), ICE1 4,249 (-0.0410), USP7 4,242 (-0.0410), SKP1 7,404 (-0.0132), ARID1A 12,593 (+0.0276), CHD4 13,245 (+0.0335), EAF1 14,482 (+0.0469), NELFE 15,340 (+0.0583), BCORL1 15,515 (+0.0609), SMARCA4 16,333 (+0.0791), RNF2 16,819 (+0.0964), HDAC1 16,942 (+0.1026), CREBBP 16,950 (+0.1032), SIN3A 17,096 (+0.1112), CDK12 17,136 (+0.1148), HDAC2 17,134 (+0.1146), EAF2 17,372 (+0.1395), PCGF1 17,405 (+0.1448), NELFA 17,650 (+0.2403), NELFB 17,671 (+0.2650). RYBP is absent from the scan.
Genome-wide KMT2A-rearranged selectivity scan, 17,728 genes (DepMap Chronos via Open Targets) · Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026
What this evidence cannot show
All rows read out of analyses/depmap-kmt2a-selective-scan-full/results.json on 2026-09-19 in a single scripted pass, so no number in this table was typed from recall. Eleven KMT2A-rearranged against 43 other leukaemia lines, AML-heavy, none infant-derived. **No gene in this table was permutation-tested, lineage-conditioned or line-normalised in this cycle**, and several of the extremes sit at the essentiality floor where the difference statistic is compressed and can invert.
Evidence that argues against this (1)
- Ranking 34 genes on one statistic from one dataset and then writing entries in rank order is a recipe for chasing artefacts. The three analyses that would discipline this - permutation, lineage conditioning and line normalisation - already exist as scripts in this repository and were not run, which is the largest methodological shortcut this beat took. NCBI PubMed, queried 2026-09-19 2026
E3moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
Okuda and Yokoyama state 'it is hypothesized that relieving the promoter proximal pausing of RNA polymerase II is the rate-limiting step of MLL fusion-dependent transcription' and report that the key activity responsible for the oncogenic property of MLL-AEP fusion proteins is TBP loading activity through SL1, and not mediator recruitment or transcriptional elongation activities.
TBP loading by AF4 through SL1 is the major rate-limiting step in MLL fusion-dependent transcription · Okuda H, Takahashi S, Takaori-Kondo A, Yokoyama A · Cell Cycle 15(20):2712-22 2016
PMID 27564129 DOI
What this evidence cannot show
Abstract fetched and read on 2026-09-19; the quoted sentence is verbatim from it. Full text not read. Murine myeloid transformation with engineered MLL-AEP constructs; one laboratory, and the AEP framework is that laboratory's own. 'Rate-limiting for transformation by constructs' is not the same as 'rate-limiting at a native fusion target gene'.
Evidence that argues against this (1)
- The pause-release model retains support: ENL, an MLL fusion partner, initiates multivalent phase separation of SEC that concentrates P-TEFb and controls pause release, and MLL-ENL fusion is reported to boost SEC phase separation. The correct summary is that the field is divided, not that pausing has been ruled out. Guo C et al. 2020
E4moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper
The ECAS complex, a bipartite assembly linking an AFF4-ELL2-EAF1 elongation module to a U2 snRNP-associated intron binding complex and distinct from P-TEFb-containing SEC, occupies promoter-proximal chromatin; acute perturbation of its AQR/XAB2-dependent functions impairs productive gene-body elongation without altering pause release and increases intron retention, while Aqr perturbation impairs leukaemia initiation, maintenance and leukaemia stem cell function in MLL-rearranged leukaemia.
A dual-module ECAS complex couples transcription elongation and RNA processing · Xiang Y, Wang H, Yin Z, Xiao R, Song Z, Lu R, Dai R, Ye J, Li C, Cui Y, Wang M, Lv J, Li Q, Wang F, Li Z, Zhang K, Zhang Y, Wu D, Fang P, Liang K · Molecular Cell, online ahead of print 16 September 2026 2026
PMID 42748909 DOI
What this evidence cannot show
Abstract fetched and read on 2026-09-19; full text not read. Very recent and unreplicated. The leukaemia phenotype is for Aqr and the pause-release observation is for AQR/XAB2 perturbation in human cells; they are separate experiments in separate systems.
Evidence that argues against this (1)
- This project's own scan does not support AQR as a KMT2A-selective dependency and points the other way (rank 17,654, difference +0.244), so the MLL-relevance of ECAS is contested inside this graph. The Okuda result is the stronger leg of the case against pause-release centrality. Broad Institute DepMap / Open Targets Platform; analyses/depmap-kmt2a-selective-scan-full 2026
E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database
Two enumeration queries in this beat had to be discarded for alias inflation: '(MLL OR KMT2A) AND (NELFE OR RDBP OR RD OR NELF-E)' returns 115 records against 2 for the corrected query without the two-letter alias RD, a 57-fold inflation; and '(MLL OR KMT2A) AND (EAF2 OR U19 OR TRAITS)' returns 75 against 5 for EAF2 alone, a 15-fold inflation.
PubMed E-utilities esearch comparison of alias-inflated and corrected queries for NELFE and EAF2, run for interactome round 4 · NCBI PubMed, queried 2026-09-19 2026
What this evidence cannot show
The inflated record sets were **not** title-checked individually; 'almost all spurious' is inferred from the magnitude of the jump and from the generic character of the tokens RD, U19 and TRAITS. It remains possible that one genuine paper sits inside an inflated set and was discarded with it, which would make a counted absence elsewhere in this beat wrong.
Evidence that argues against this (1)
- The standing brief's rule points the other way and is also right: for the septin partners, dropping the retired symbol lost nine-tenths of the literature. Recall and precision pull in opposite directions here and neither rule can be applied mechanically. The only safe procedure is to title-check the difference set whenever adding an alias changes a count by more than a few records. NCBI PubMed, queried 2026-09-19; finding originally made by this project's non-elongation-partners beat, cycle 40 2026
E6strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper
Of the 7 records returned by '(ICE1 OR ICE2) AND "little elongation complex"', at least three name ICE1 only incidentally and study something else: a targeted resequencing study of a heparin-induced thrombocytopenia GWAS locus (PMID 29934777), a genome-wide parent-of-origin-by-environment analysis of cleft palate trios (PMID 28898263), and a proteomic comparison of skeletal-muscle extracellular vesicles from young and old donors (PMID 39169919).
Targeted resequencing of a locus for heparin-induced thrombocytopenia on chromosome 5 identified in a genome-wide association study · Witten A, Bolbrinker J, Barysenka A, Huber M, Ruhle F, Nowak-Gottl U, Garbe E, Kreutz R, Stoll M · Journal of Molecular Medicine (Berlin) 96(8):765-775 2018
PMID 29934777 DOI
What this evidence cannot show
All three abstracts fetched and read on 2026-09-19 to confirm they are not LEC studies. Only one is cited as representative. This evidence item documents an enumeration hazard and supports no biological claim.
Evidence that argues against this (1)
- Judging a record 'incidental' is a call that can be wrong. A heparin-induced thrombocytopenia locus is a haematological phenotype and a gene appearing there is not self-evidently irrelevant to haematopoiesis - the narrower and defensible claim is only that these papers are not about LEC or leukaemia. Fernandez-Rhodes M et al. 2024

Connected to

supports →PCGF1 — the PRC1.1 scaffold whose LOSS causes menin-inhibitor resistance, so it is a restraint on the leukaemia rather than a dependency of it
enumeration ledger recording the counts and rejections behind every entry in this beat
supports →NELF — the brake that has to come off for elongation, absent from the KMT2A literature, and less required in KMT2A-rearranged lines than in any other leukaemia
documents the counted absence of NELF in KMT2A-rearranged literature and the granularity exception for NELFA and NELFCD
supports →SUPT5H (SPT5) — the pausing factor the KMT2A field has never studied, and the counted absence that undercuts 'the fusion hijacks pause release'
documents the DSIF counted absence and the two published results placing the rate-limiting step away from pause release
supports →KMT2A-rearranged cells need the factors retained through elongation MORE and the factors ejected at pause release LESS — a six-gene asymmetry nobody has reported
the genome-wide dependency read-out from which the asymmetry was derived
supports →NELFE (RDBP) — the RNA-binding subunit of the pausing brake, and the alias that inflated a record count fifty-fold before it was caught
the RD alias inflation (115 records against 2) that this beat caught and discarded
supports →EAF2 (U19) — the EAF paralogue whose ELL-binding domain is a mapped tumour-suppressor module, described entirely in prostate cancer
the U19/TRAITS alias inflation (75 records against 5)
supports →ICE1 — the 2,266-residue scaffold of the little elongation complex, the subunit that acts at initiation, and zero papers connecting it to leukaemia of any kind
three of seven little-elongation-complex records shown to be incidental mentions in unrelated studies
supports →BRD9 — dispensable in the average cell, rank 39 of 17,728 in KMT2A-rearranged lines, and with an on-target proof most epigenetic targets never get
BRD9, MYB, CEBPA and SPI1 emerged from the genome-wide scan rather than from the candidate list the brief supplied
supports →PRC1.1 — the complex this graph has been leaning on, now enumerated subunit by subunit, and the enumeration changes the conclusion
subunit-level counts and dependency rows behind the complex-level PRC1.1 synthesis
← related toInteractome round 5: 42 queries, 13 entries written, 9 candidates rejected with counts, and four alias failures of four different kinds
Reports two corrections to round 4's rejection table - SETD1B and CXXC1 - neither changing a number and both changing a conclusion.
← refutesTwo of the three top new dependencies are myeloid identity, not KMT2A
Two of the three myeloid transcription factors this enumeration nominated as top dependencies do not survive a lineage split of the index panel.
← related toA real defect in a committed scan, computed three days ago and never written up: `depmap-kmt2a-selective-scan-full` fixes both comparison arms from the FIRST gene that returns data, so five of sixteen eligible KMT2A-rearranged lines are excluded from all 17,728 gene comparisons - including the genes where they were screened
one of six entries whose data.script is the affected scan
← related toMTA2 / MTA1L1 (O94776) - the NuRD scaffold that clamps HDAC1, recruited in T lymphocytes by BCL11B, and absent together with MTA1 and MTA3 from every column of the pan-MLL-fusion interactome
this entry confirms rather than overturns that round's rejection of MTA2 (0 records, rank 1,037 inside noise)