Oncofusion

start / graph / gene-actb

Geneweak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.Q90-pan-fusion-core3

ACTB / beta-actin (P60709) as a SUBJECT, not a loading control - the dirtiest negative control in the pan-MLL-fusion 42, and the one the paper's authors kept

also called ACTB · beta-actin · Actin, cytoplasmic 1 · ACTB_HUMAN · P60709

Beta-actin is the protein that builds the cell's internal cables. It is so abundant that biologists routinely use it as a background reference on a gel - which is exactly why finding it stuck to a cancer protein is suspicious. But actin also genuinely works inside the nucleus, where it is a built-in part of the machines that open up DNA, so this one cannot simply be dismissed.

The same thing, for a scientist

READ THE FIRST LINE BEFORE USING THIS ENTRY. `data/symbol-collisions.json` carries an ACTB row classed `token_is_a_reagent_not_a_subject` with `block_autogeneration: true` and the note that ACTB 'should NOT be written' - because the token appears in 46 entries of this graph as a qPCR amplicon or a DepMap negative-control row. That block is correct and this entry does not overturn it. It blocks writing ACTB off a token census. THIS entry writes ACTB as a SUBJECT, on one specific published measurement: beta-actin (P60709, 375 aa) is one of the 42 proteins recovered by all seven MLL-fusion baits in Skucha 2018, MLL-AF4 p = 0.0145 with spectral counts 22|16, and it carries a **3|3 negative control - the dirtiest of the 42**, with PSIP1/LEDGF second at 0|5. Three measurements made here decide how to read it. (1) Over all 960 rows, neg_ctrl total correlates with maximum per-bait spectral abundance at Spearman +0.461, so ACTB's dirty control is largely a statement that it is the most abundant prey in the set - it reaches 388 spectra under MLL-CBP. (2) ACTB survived the authors' manual removal of seven contaminants and entered the 128-gene shRNA screen - but so did CCT3, and CCT/TRiC is the canonical AP-MS contaminant which this project's own complex analysis uses as its NEGATIVE control, so surviving that filter certifies nothing. (3) The strongest pro-interaction argument - that beta-actin is a stoichiometric subunit of the BAF/SWI-SNF complex, whose subunits SMARCB1, SMARCC1 and SMARCC2 are all in the MLL-AF4 column - has a measurable hole: **ACTL6A/BAF53A, beta-actin's obligate partner in the BAF ARP module, is completely absent from the 960-protein network.** Verdict: genuinely undecidable, and the most interesting of the nine.

Go deeper
Why this entry exists despite the registry block

The registry is right that ACTB in this corpus is usually a reagent: a TREC/KREC/ACTB qPCR amplicon, a DepMap negative-control row, a loading control on a blot. Writing gene-actb off a count of those 46 mentions would be a category error of the same family as writing a gene entry for CD15. The block is against autogeneration from a token census, and this entry is not that. Skucha 2018 Supplementary Table 5 lists ACTB as a prey with an accession, a p-value, spectral counts and a negative control, recovered by all seven baits, and carried by the authors into a functional screen. That is a subject. Having the entry also does something the block cannot: it gives the graph one place that states which register is which. Correction owed outside this file set: the ACTB registry row's note should be amended from 'should NOT be written' to 'should not be written FROM THE TOKEN QUEUE; one subject-use exists, Skucha 2018 Supplementary Table 5, and it has an entry at gene-actb.'

The Skucha row, stated in full

7 of 7 fusions. MLL-AF1p 0.00772 (18|22), MLL-AF4 0.0145 (22|16), MLL-AF9 0.0407 (34|26), MLL-CBP 4.38e-07 (41|347), MLL-EEN 0.0697 (31|42), MLL-ENL 2.59e-21 (183|133), MLL-GAS7 0.0407 (47|31); negative control 3|3, the highest of the 42 1. Rank 20 of 42 by MLL-AF4 p. Two things to notice. Its MLL-EEN p is 0.0697 - above 0.05 even though it is in the pan-fusion set, which is the rank-cutoff property in one row. And MLL-CBP's replicates are 41 and 347, an 8.5-fold discordance between two replicates of the same purification; nothing else in this tranche does that, and it is the shape carryover makes.

The strongest argument that it is real, and the measurement that undercuts it

Beta-actin is not merely a nuclear passenger. Purification and peptide sequencing of the mammalian BAF (SWI/SNF-like) complex from T lymphocytes identified beta-actin and the novel actin-related protein BAF53 as subunits, both required for maximal BRG1 ATPase activity and for the complex's association with chromatin and matrix 2. Nuclear actin polymerisation is separately required for DNA double-strand break clustering in homology-directed repair, with WASP and ARP2/3 recruited to damaged chromatin 3. UniProt's human record carries 'Nucleus' on two experimental citations. So a nuclear pool of beta-actin with chromatin-complex membership is established, in a lymphocyte system, which is the right lineage for a Jurkat experiment.

And here is the hole, measured by this beat against the full 960-row table on 2026-09-21: ACTL6A/BAF53A is completely absent from the network at any n. BAF53 and beta-actin are the two halves of the BAF ARP module and were co-purified in the same 1998 experiment. The BAF subunits that ARE present are the core scaffold - SMARCB1 (n = 5, MLL-AF4 p 0.0178), SMARCC1 (n = 7, 0.0707), SMARCC2 (n = 3, 0.0863) - with SMARCA4 at n = 1 and ARID1A, SMARCA2 and SMARCD1 absent entirely. If ACTB were arriving as a BAF subunit, ACTL6A should have arrived with it. It did not 12.

Surviving the authors' contaminant filter is worth less than it sounds

The Methods state, verbatim: 'The proteins shared by at least five baits were selected for further analysis. Seven proteins were manually removed as these were either frequently observed contaminants or were not detected in human hematopoietic cell lines. The final set was comprised of 128 proteins.' Measured by this beat on 2026-09-21: the set of 128 symbols in the shRNA library (MOESM5) is identical to the 128 rows of MOESM4 with n_fusions >= 5, the only differences being four alias spellings. So the seven removed proteins are not recoverable from the published table - and ACTB is in the surviving 128 and was screened with six shRNAs.

But so is CCT3. The CCT/TRiC chaperonin is the canonical AP-MS contaminant; this project's own analyses/mll-fusion-interactome-complexes/ uses it as the NEGATIVE control and it is still significant against a realistic nuclear background (q = 2.7e-05, 8 of 9 subunits in the MLL-AF4 column, 6 at p < 0.05). The authors' own named essentiality controls were RFC1, SF3A3 and CCT3. A filter that let a whole canonical contaminant complex through does not certify beta-actin 1.

Nothing outside this experiment ties ACTB to a KMT2A fusion

ACTB[tiab] AND (KMT2A[tiab] OR MLL[tiab] OR "MLL-AF4"[tiab] OR "MLL-AF9"[tiab] OR "t(4;11)"[tiab]) returns 4, and all four were title-checked and are off-topic: a hypertension twin epigenomics study, two neurodevelopmental exome papers, and a cognitive-genetics review - ACTB appearing as a gene in a variant list or as a normalisation gene. Zero on topic. The identical clause with MEN1 returns 85.

Verdict

Genuinely undecidable, and the most interesting of the nine. Everything that would make it a contaminant is present (highest abundance in the set, dirtiest control, an 8.5-fold replicate discordance under one bait, no independent KMT2A tie) and everything that would make it real is also present (an experimentally documented nuclear pool, membership in a chromatin remodelling complex whose scaffold subunits are in the same column, the right lineage, and survival of the authors' manual filter). The single fact that most cuts against it is the one nobody had looked up: ACTL6A is not in the network.

Facts and figures
uniprotP60709
length AA375
accession resolutionResolved live 2026-09-21: `gene_exact:ACTB AND organism_id:9606 AND reviewed:true` returns exactly ONE reviewed row; gene_primary ACTB, P60709, entry name ACTB_HUMAN, 375 aa, protein name 'Actin, cytoplasmic 1 (EC 3.6.4.-) (Beta-actin) [Cleaved into: Actin, cytoplasmic 1, N-terminally processed]', no gene synonyms. Matches the accession in Skucha's own table.
identifier hazardREGISTERED AND BLOCKED, read BEFORE searching per rule 88. `data/symbol-collisions.json` row: token ACTB, class `token_is_a_reagent_not_a_subject`, `block_autogeneration: true`, named_in_entries 46, note 'Ranked high on the named-but-unwritten queue and should NOT be written. Every match checked is a loading or negative control.' THAT BLOCK IS CORRECT FOR THE TOKEN QUEUE AND THIS ENTRY DOES NOT OVERTURN IT. This entry is written from a single named subject-use - Skucha 2018 Supplementary Table 5 - not from the token census. An amendment to the registry note is owed and is recorded in the memo; this entry does not edit that file because another agent is writing concurrently. SECOND HAZARD, measured here: beta- and gamma-cytoplasmic actin (ACTB P60709 / ACTG1 P63261) differ by four N-terminal residues and share essentially all tryptic peptides, so ACTG1's absence from the 960 is a protein-inference artefact and must not be read as beta-actin specificity. Length is no defence in this family (375 vs 375).
localizationCYTOPLASM AND NUCLEUS, both experimental. UniProt P60709, fetched live 2026-09-21 with per-location evidence codes read individually: 'Cytoplasm, cytoskeleton' ECO:0000269 PubMed:11687588 and 17289661; 'Nucleus' ECO:0000269 PubMed:11687588 and 29925947. Both nuclear citations were fetched and read by this beat rather than taken on UniProt's pointer (rule 91): 11687588 reports nuclear F-actin binding NDH II/RNA helicase A in HeLa nuclear extract with electron-microscopic colocalisation, and 29925947 reports nuclear actin and ARP2/3 at DNA double-strand breaks. Both are genuinely about nuclear actin, so UniProt's pointer is sound here - which is not the default in this graph's experience.
subnuclear localizationNO SPECIFIC COMPARTMENT ASSERTED, and this is a genuine gap rather than a not-applicable. Beta-actin has an experimentally supported nuclear pool, so a subnuclear question is meaningful for it, unlike the purely cytoplasmic members of this tranche. What is documented is functional rather than compartmental: actin polymers at DNA double-strand break clusters in G2 (PMID 29925947) and association with the nuclear matrix as part of BAF (PMID 9845365). UniProt records no subnuclear compartment. Nobody has imaged nuclear actin in a KMT2A-rearranged cell.
localization shown for fusionNO - nobody has imaged or fractionated beta-actin in any cell expressing any KMT2A fusion.
shown for fusionCo-purification with all seven MLL fusions including MLL-AF4 in Jurkat (p = 0.0145, neg_ctrl 3|3); shRNA-screened against MLL-AF9 in MOLM-13, score unpublished. NO independent experiment ties beta-actin to any KMT2A fusion - the bounded query returns 4 records, all title-checked off-topic.
selectivityUNKNOWN, and BARRED from any target list under any circumstance. Beta-actin is pan-essential; DepMap-style selectivity is meaningless for it and this graph already records ACTB as a DepMap NEGATIVE-CONTROL row. Not named among the paper's non-selectivity controls (RFC1, SF3A3, CCT3) and not known to be among the 40 re-screened. The paper's own precedent is instructive: SF3A3 is a named non-selectivity control and is in this same pan-fusion 42, so pan-fusion membership and therapeutic irrelevance are fully compatible.
absence classnever-measured
absence class noteNo instrument has been pointed at beta-actin's association with a KMT2A fusion in any cell. The MOLM-13 shRNA score is measured-not-reported (and will be uninterpretable - ACTB is pan-essential, so every shRNA will deplete). PRIDE checked: PXD009338 is the only relevant deposition and is the source experiment.
graph contextThere is no `gene-actb` node before this entry; the token appears in 46 entries, every checked instance as a reagent (qPCR amplicon, loading control, DepMap negative-control row), which is what the registry row records.
verdictUNDECIDABLE, and the most interesting of the nine. Contaminant evidence: highest abundance in the set, dirtiest negative control of the 42, 8.5-fold replicate discordance under MLL-CBP, MLL-EEN p above 0.05, no independent KMT2A tie, and a CRAPome-class background profile. Real-interaction evidence: an experimentally documented nuclear pool on two independent papers, stoichiometric membership of the BAF chromatin remodelling complex whose scaffold subunits are in the same column, and the right lineage. The decisive new measurement is that ACTL6A - beta-actin's obligate BAF partner - is absent from the network, which removes the best mechanistic story.
skucha 2018
{
  "bait": "MLL-AF4 (= KMT2A::AFF1)",
  "p_value": 0.0145,
  "spectral_counts": "22|16",
  "n_fusions_of_seven": 7,
  "negative_control_counts": "3|3",
  "negative_control_rank": "1 of 42 - the dirtiest negative control in the pan-fusion set. Second PSIP1 0|5, third CHD4 2|2, then SF3A3 0|3. Ten of the 42 are non-zero.",
  "per_bait": {
    "MLL-AF1p": "0.00772 (18|22)",
    "MLL-AF4": "0.0145 (22|16)",
    "MLL-AF9": "0.0407 (34|26)",
    "MLL-CBP": "4.38e-07 (41|347)",
    "MLL-EEN": "0.0697 (31|42)",
    "MLL-ENL": "2.59e-21 (183|133)",
    "MLL-GAS7": "0.0407 (47|31)"
  },
  "af4_spectral_rank_of_seven": 7,
  "af4_p_rank_within_the_42_pan_fusion": 20,
  "replicate_discordance": "MLL-CBP 41|347, an 8.5-fold discordance between two replicates of one purification; unique in this tranche and the shape carryover produces. MLL-ENL is 183|133, concordant but enormous.",
  "source": "analyses/skucha-mll-fusion-interactome/results.json, re-derived from the cached MOESM4 xlsx 2026-09-21; run.py re-run, both controls PASS"
}
complex membership
{
  "complex": "BAF / SWI-SNF (beta-actin is a documented subunit of the ARP module, with ACTL6A/BAF53A)",
  "partners_in_the_960": {
    "SMARCB1": "n=5, MLL-AF4 p 0.0178 (4|4), neg 0|0",
    "SMARCC1": "n=7, MLL-AF4 p 0.0707 (5|4), neg 1|0",
    "SMARCC2": "n=3, MLL-AF4 p 0.0863 (5|1), neg 0|0",
    "SMARCA4": "n=1, no MLL-AF4 value, neg 0|1",
    "SMARCE1": "n=1, no MLL-AF4 value, neg 0|0"
  },
  "absent_from_the_960": [
    "ACTL6A (BAF53A) - the decisive absence",
    "ACTL6B",
    "ARID1A",
    "SMARCA2",
    "SMARCD1",
    "ACTR5",
    "ACTR8",
    "ACTR6",
    "INO80",
    "RUVBL1",
    "RUVBL2",
    "EP400",
    "KAT5",
    "ACTG1 (but see identifier_hazard)"
  ],
  "reading": "PARTIAL SCAFFOLD, NO ARP MODULE. The BAF core scaffold is present at low n; beta-actin's obligate ARP-module partner ACTL6A is absent, as is every other actin-related protein in the cell's chromatin machinery (ARP5, ARP8, ARP6, the INO80 and NuA4/TIP60 ATPases). Against a realistic 5,754-protein nuclear background, analyses/mll-fusion-interactome-complexes/ scores BAF/SWI-SNF at q = 0.076, NOT significant - one of the seven of twelve complexes that stop being significant once the background is corrected. So there is no secure BAF module in this dataset for ACTB to belong to."
}
cell types and states
[
  "Jurkat (human T-ALL) - the AP-MS host, nuclear lysate, tagged single-copy transgene",
  "MOLM-13 (human MLL-AF9 AML) - shRNA screen host, six shRNAs, per-gene score unpublished",
  "primary and transformed T lymphocytes - the BAF purification that identified beta-actin as a subunit",
  "Xenopus egg extract and mammalian cells - the nuclear ARP2/3 DSB work",
  "no B-lineage cell, no infant cell, and no KMT2A::AFF1 cell of any kind"
]
functionally screened in the source paper
{
  "screened": true,
  "shrnas_for_this_gene": 6,
  "design": "Supplementary Table 6 / MOESM5 is the shRNA library for the negative-selection screen: 768 data rows, 128 unique gene symbols, EXACTLY 6 shRNAs each - counted with openpyxl by this beat on 2026-09-21. Host is MLL-AF9-expressing MOLM-13, a human myeloid line.",
  "set_identity_measured_by_this_beat": "The 128 symbols in the shRNA library are the SAME SET as the 128 rows of MOESM4 with n_fusions >= 5, checked by set difference on 2026-09-21: the only four mismatches are alias spellings (JADE2/PHF15, JADE3/PHF16, KMT2A/MLL, NIFK/MKI67IP). This matters for the contamination question. The paper says 'Seven proteins were manually removed as these were either frequently observed contaminants or were not detected in human hematopoietic cell lines. The final set was comprised of 128 proteins.' Because the table's >=5 set is already exactly 128, those seven are NOT recoverable from Supplementary Table 5 - and every protein in this tranche is in the post-removal 128 and was carried into the functional screen.",
  "what_this_does_NOT_establish": "A per-gene result. Scores are presented only as heatmaps (Fig. 2c/2d); no per-gene depletion value is tabulated anywhere in the paper or its supplementary files, so for any individual gene the outcome is measured-not-reported. It is also MLL-AF9 MOLM-13, a myeloid line, not the index fusion and not B-lineage.",
  "the_filter_is_leaky_and_here_is_the_proof": "'Survived the authors' manual contaminant removal' is weaker than it sounds. CCT3 is also in the 128 and was also screened, and the CCT/TRiC chaperonin is the canonical AP-MS contaminant - the companion analysis analyses/mll-fusion-interactome-complexes/ uses it as its NEGATIVE control and it is still significant against a nuclear background (q = 2.7e-05, 8 of 9 subunits in the MLL-AF4 column). The authors' own named essentiality controls were RFC1, SF3A3 and CCT3. So the manual removal did not remove a canonical contaminant complex, and membership in the 128 is not the authors certifying a protein as a real partner.",
  "absence_class_for_the_per_gene_score": "measured-not-reported",
  "library_symbol": "ACTB"
}
bounded literature
{
  "query": "ACTB[tiab] AND (KMT2A[tiab] OR MLL[tiab] OR \"MLL-AF4\"[tiab] OR \"MLL-AF9\"[tiab] OR \"t(4;11)\"[tiab])",
  "querytranslation": "\"ACTB\"[Title/Abstract] AND (\"KMT2A\"[Title/Abstract] OR \"MLL\"[Title/Abstract] OR \"MLL-AF4\"[Title/Abstract] OR \"MLL-AF9\"[Title/Abstract] OR \"t 4 11\"[Title/Abstract])",
  "count": 4,
  "on_topic": 0,
  "what_the_four_are": "Title-checked individually 2026-09-21: 39972178 (epigenomics of blood pressure in twins), 34948243 (autism/ID whole-exome candidates), 33659785 (longitudinal genetics of cognition), 29158550 (Exome Pool-Seq in neurodevelopmental disorders). ACTB appears in each as a variant-list gene or a normalisation gene; none concerns KMT2A.",
  "date": "2026-09-21",
  "tool": "NCBI esearch db=pubmed",
  "names_tried": [
    "ACTB",
    "beta-actin (not run scoped - the phrase is a reagent term in tens of thousands of records and would not have been interpretable)"
  ],
  "positive_control": "MEN1[tiab] AND the identical five-term KMT2A clause returns 85.",
  "bound": "Titles and abstracts only; supporting figures not searched (rule 80). Cytogenetic term quoted, querytranslation read back with no injected [UID] term (rule 95). One tag per term (rule 94). NOTE this is a case where a literature bound is weak in principle: ACTB is a reagent term, so its absence from titles and abstracts in a KMT2A context is close to guaranteed regardless of the biology."
}
what would settle it
[
  "Reanalyse PXD009338 for ACTL6A peptides below the 300-per-bait rank cutoff. If BAF53A is there sub-threshold, the BAF-subunit reading revives; if it is genuinely absent, ACTB is not arriving as a BAF subunit.",
  "Retrieve the CRAPome frequency for P60709. Free, and it is the external background control the source paper did not run.",
  "Ask the authors why MLL-CBP replicate 2 gave 347 spectra against replicate 1's 41. A within-purification answer settles more than any new experiment."
]
How this could bite us back
  • ACTB can never be a target. It is pan-essential, structural, and this graph already carries it as a DepMap negative control; any depletion phenotype is an assay artefact, not a therapeutic signal.
  • The real bite-back is epistemic: this entry is the graph's statement that a protein can be a reagent in 46 places and a subject in one, and that the registry block protects the first without forbidding the second. If a future pass reads only the registry row it will conclude this entry should not exist.
  • If nuclear actin does turn out to matter here, the intervention space is worse than usual - the available tools (latrunculin, cytochalasin, CK-666) hit cytoplasmic actin dynamics first and would kill any cell long before a nuclear-specific effect could be read out.
Still open
  1. Is ACTL6A/BAF53A present in PXD009338 below the rank cutoff? This one reanalysis decides whether beta-actin's recovery has a BAF explanation, and it needs no wet work.
  2. What explains MLL-CBP's 41 vs 347 replicate split? No other protein in this tranche behaves that way, and an answer would calibrate how much replicate discordance the whole table carries.
  3. Does nuclear actin polymerisation matter in KMT2A-rearranged leukaemia at all? The DSB-clustering role intersects this graph's SETD2/DNA-damage axis - Skucha's own headline is that SETD2 loss increases DNA damage - and nobody has connected them.
  4. Which seven proteins did the authors manually remove? They are not recoverable from Supplementary Table 5, and knowing them would calibrate every contamination judgement in this tranche. That is one email.
Every source behind this entry (3)
  1. E1weak confidence 2 of 5: Preliminary, small, indirect, or not yet peer reviewed.paper

    Beta-actin co-purified with all seven Strep-HA-tagged MLL fusions from nuclear lysates of isogenic human Jurkat T-ALL cells; for MLL-AF4 (= KMT2A::AFF1) the combined p-value is 0.0145 with spectral counts 22|16 and a negative control of 3|3, the highest negative control of the 42 pan-fusion proteins; it is in the paper's final post-filter set of 128 and was screened with six shRNAs in MLL-AF9 MOLM-13.

    MLL-fusion-driven leukemia requires SETD2 to safeguard genomic integrity. · Skucha A, Ebner J, Schmoellerl J, Roth M, Eder T, Cesar-Razquin A, Stukalov A, Vittori S, Muhar M, Lu B, Aichinger M, Jude J, Mueller AC, Gyorffy B, Vakoc CR, Valent P, Bennett KL, Zuber J, Superti-Furga G, Grebien F · Nature Communications 2018
    PMID 29777171DOI
    What this evidence cannot show

    SYSTEM: the AP-MS host is JURKAT, a human T-ALL line, while KMT2A::AFF1 disease is B-lineage infant leukaemia; a partner requiring a B-cell or infant proteome cannot appear here and a T-lineage partner can appear spuriously. The bait is a Strep-HA-tagged, SINGLE-COPY, doxycycline-induced transgene at an FRT/Flp cassette, not the endogenous fusion. Prey came from bulk nuclear lysate after a single STREP-Tactin step, so co-purification is NOT evidence of a direct contact and is not evidence that the two proteins are ever in the same complex. INCLUSION RULE: the paper kept a fixed number of top-ranked interactions per bait rather than applying a p threshold. Methods, read verbatim from the PMC full text by this beat on 2026-09-21: "All the identified bait-prey pairs were ranked by the combined p-value. The 300 most significant interactions per bait were retained." Measured by this beat from the cached MOESM4 xlsx: within the MLL-AF4 column 186 of 291 listed p-values (63.9%) exceed 0.05, and that column's maximum p is 0.175. BAIT DEPTH: for all nine proteins in this tranche the MLL-AF4 spectral total is rank 7 of the seven baits - measured protein by protein - so a low MLL-AF4 count is a bait property (transgene expression, purification depth, or the GLM normalisation) and must not be read as a weaker association than the same protein's count under another bait. HOW TO READ THE NEGATIVE CONTROL, measured by this beat on 2026-09-21 over all 960 rows of MOESM4: the Spearman correlation between a protein's neg_ctrl spectral total and its maximum per-bait spectral total is +0.461, and the median maximum per-bait spectral count is 42 for the 145 proteins (15.1%) with a non-zero negative control against 8 for the 815 with 0|0. A dirty negative control in this table is therefore substantially a statement about protein ABUNDANCE, not about specificity, and a clean 0|0 is substantially a statement that two control replicates were not deep enough to see the protein at all. Two consequences. (1) PSIP1/LEDGF - the one pan-fusion partner with a solved co-crystal against the KMT2A N-terminus - carries 0|5, the second-dirtiest control of the 42, so a non-zero control cannot be used to call a protein a contaminant. (2) A 0|0 control is not purity; it is an unestimated baseline. The paper's GLM estimates each prey's baseline abundance A0,j from the control purifications, so a prey the controls never saw has its baseline set by a floor rather than by data, which is the direction that inflates apparent specificity. WHERE THIS ROW SITS AMONG THE 42, measured by this beat on 2026-09-21: ranked by MLL-AF4 p-value the pan-fusion set runs from MEN1 at 9.75e-16 to SMC3 at 0.174. The three WEAKEST MLL-AF4 calls in the whole pan-fusion set - SMC3 0.174, NOLC1 0.134, TRIM28 0.129 - all already have entries in this graph, and gene-nolc1 and gene-trim28 both carry confidence_overall 'weak' with those p-values recorded. So p > 0.05 in the index-fusion column is not a disqualifier in this corpus; it is a confidence tier that two existing entries already occupy. SPECIFIC TO THIS PROTEIN, all measured by this beat from the cached MOESM4 xlsx on 2026-09-21. (a) ACTB's neg_ctrl of 3|3 is the highest of the 42 pan-fusion proteins; PSIP1/LEDGF is second at 0|5 and CHD4 third at 2|2, and all three are proteins this graph treats as real, so a dirty control does not discriminate. (b) ACTB is the most abundant prey in the set, reaching 388 spectra under MLL-CBP, which by the +0.461 abundance correlation is sufficient to explain the dirty control on its own. (c) MLL-CBP's two replicates are 41 and 347 - an 8.5-fold discordance within one purification, unique in this tranche and the shape carryover produces. (d) ACTB's MLL-EEN p is 0.0697, above 0.05, inside the pan-fusion set. (e) ACTG1 is absent from the 960, which must NOT be read as beta-actin specificity: beta- and gamma-cytoplasmic actin differ by four residues at the N-terminus and share essentially all tryptic peptides, so a protein-inference step assigns the shared evidence to one entry. SELECTIVITY: not named among the paper's controls (RFC1, SF3A3, CCT3 non-selective; MEN1 selective); not known to be among the unpublished 40 re-screened. Recorded as unknown. run.py re-run 2026-09-21, both controls PASS.

    Evidence that argues against this (2)
    • Surviving the authors' manual contaminant removal certifies nothing, because CCT3 survived it too. The CCT/TRiC chaperonin is the canonical AP-MS contaminant and is used as the explicit NEGATIVE control in this project's own complex analysis, where it remains significant against a 5,754-protein nuclear background at q = 2.7e-05 with 8 of 9 subunits in the MLL-AF4 column. Measured here on 2026-09-21: the 128-symbol shRNA library and the 128 table rows with n_fusions >= 5 are the same set, so the seven removed proteins cannot be recovered and the filter's stringency cannot be audited. ProteomeXchange/PRIDE 2018
    • Actin is among the most frequently observed background proteins across published affinity purifications, and the CRAPome authors state that small-scale studies with limited negative controls do not capture a complete background protein set. Skucha 2018 scored against its own control purifications only and consulted no contaminant repository. A CRAPome frequency for P60709 was not retrieved by this beat and is the cheapest external check available on this row. Mellacheruvu D et al. 2013
  2. E2established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Purification and peptide sequencing of the mammalian SWI/SNF-like BAF complex from T lymphocytes identified beta-actin and a novel actin-related protein, BAF53, as subunits of the complex; both are required for maximal ATPase activity of BRG1 and, with BRG1, for association of the complex with chromatin and matrix.

    PMID 9845365DOI
    What this evidence cannot show

    1998; biochemical purification plus peptide sequencing from T lymphocytes, with in vitro ATPase and chromatin/matrix association assays. This establishes that beta-actin is a genuine nuclear chromatin-complex subunit - it does NOT establish any contact with KMT2A or with an MLL fusion, and BAF is not the MLL complex. Read from the abstract. The lineage is the relevant strength here: the BAF work was done in T lymphocytes and the Skucha AP-MS host is Jurkat, a T-ALL line.

    Evidence that argues against this (1)
    • The argument this citation supports has a measurable hole. Beta-actin and BAF53 (ACTL6A) were co-purified as the two halves of the BAF ARP module in this very paper - and ACTL6A is COMPLETELY ABSENT from Skucha's 960-protein network at any n, checked by this beat against the full table on 2026-09-21. The BAF subunits recovered are the scaffold - SMARCB1 (n=5, MLL-AF4 p 0.0178), SMARCC1 (n=7, 0.0707), SMARCC2 (n=3, 0.0863) - with SMARCA4 at n=1 and ARID1A, SMARCA2 and SMARCD1 absent. If ACTB arrived as a BAF subunit its obligate partner should have arrived with it. Separately, the project's own complex analysis finds BAF/SWI-SNF is NOT significant against a realistic nuclear background (q = 0.076 ns; only 6 of 17 subunits found), so even the scaffold recovery is not a secure module call. Skucha A et al. 2018
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.paper

    Nuclear actin, WASP and the actin-nucleating ARP2/3 complex are recruited to damaged chromatin undergoing homology-directed repair; nuclear actin polymerisation is required for the migration of a subset of double-strand breaks into discrete sub-nuclear clusters, and inhibiting actin nucleation impairs DNA end-processing and homology-directed repair.

    Nuclear ARP2/3 drives DNA break clustering for homology-directed repair. · Schrank BR, Aparicio T, Li Y, Chang W, Chait BT, Gundersen GG, Gottesman ME, Gautier J · Nature 2018
    PMID 29925947DOI
    What this evidence cannot show

    Xenopus laevis cell-free extracts plus mammalian cells; not haematopoietic, not a fusion system. This is one of the two experimental citations behind UniProt P60709's 'Nucleus' location comment (the other is PubMed 11687588, nuclear F-actin binding NDH II/RNA helicase A in HeLa nuclear extract) - both were fetched and read by this beat on 2026-09-21 rather than taken from UniProt's pointer, per rule 91, and both genuinely concern nuclear actin. It establishes a functional nuclear actin pool; it says nothing about KMT2A. Read from the abstract.

    Evidence that argues against this (1)
    • A nuclear actin pool makes the Skucha row possible without making it specific: everything in a nuclear lysate is in the same tube as the bait, and a protein that is genuinely nuclear AND the most abundant prey in the dataset is the hardest case for an AP-MS background model. The same logic that rescues ACTB from 'it cannot be in a nucleus' also removes the only clean reason to exclude it, which is why this entry's verdict is undecidable rather than positive. Mellacheruvu D et al. 2013
Connected to