Oncofusion

start / graph / gene-dna2

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q32-brca1-hr-licensing

DNA2 (P51530, 1,060 aa) - EXO1's redundant partner in long-range resection, dual mitochondrial and nuclear, and the gene whose DepMap row is measured in 80 screens where every other cascade member has 1,258

also called DNA2 · DNA2L · KIAA0083 · P51530 · DNA2_HUMAN · DNA replication ATP-dependent helicase/nuclease DNA2 · DNA replication ATP-dependent helicase-like homolog

DNA2 is the second of the two enzymes that trim back a broken chromosome end far enough for accurate repair, and it works in the mitochondria as well as the nucleus. Its dependency data looks dramatic but is measured in sixteen times fewer cell lines than every other gene in this cascade, so the number cannot be compared with theirs.

The same thing, for a scientist

**Identity.** P51530, `DNA2_HUMAN`, **1,060 aa**, primary gene name DNA2, UniProt gene synonyms **DNA2L and KIAA0083**, recommended name *DNA replication ATP-dependent helicase/nuclease DNA2*, **one** PDB entry (5EAY). A UPF1-type SF1B helicase ATP-binding domain 625-823 and its C-terminal domain 824-1052; regions typed for nuclease activity 81-519 and helicase activity 520-1060 [E1]. One reviewed record; `gene_primary` reads back DNA2. **Position in the cascade.** DNA2 appears in [E2] only as the second half of the pair 'EXO1/DNA2' - the long-range resection effectors whose recruitment is abrogated when the CTCF-directed repressive-chromatin assembly fails. The paper does not separate them, and neither should this entry: every statement it makes about DNA2 is a statement about the pair. **Two things about DNA2 are its own, and both matter.** 1. **It is a dual-compartment protein.** UniProt records both Nucleus and **Mitochondrion**, with the uncoded note that DNA2 *'was initially reported to be exclusively mitochondrial (PubMed:18995831); however, it was later shown to localize both in mitochondrion and nucleus (PubMed:19487465)'*. Both assignments are **uncoded** - there is no ECO:0000269 on either. So the compartment split rests on a curator's summary of a correction, and any DNA2 phenotype has a mitochondrial explanation available that EXO1's does not. 2. **Its DepMap row is a different measurement from everyone else's.** In one call, one data version, DNA2 returns **80 screens across 17 tissue blocks** while BRCA1, BARD1, CBX3, CBX5, KDM5A, EXO1 and CTCF each return **1,258 across 29**. Its mean gene effect is **-2.71** with **100%** of lines below -1.0 [E3]. Rule 83 says gene-specific denominators are legitimate and most such spread is real - but a sixteen-fold difference alongside a mean four times deeper than any other cascade member is a signal to state and not to rank on. **This entry does not claim DNA2 is a deeper dependency than EXO1.** It records two numbers from two denominators and refuses the comparison. **Literature at this graph's locus: empty, with large operands.** `DNA2[tiab] AND KMT2A[tiab]` = 0 and `DNA2[tiab] AND MLL[tiab]` = 0, both with non-empty querytranslations and warninglist 'No items found.'; bare `DNA2[tiab]` = **709**, `DNA2[tiab] AND leukemia[tiab]` = **1** [E4]. **Archive: closed.** ENCODE has no DNA2 target record [E5].

What it looks like
DNA replication ATP-dependent helicase/nuclease DNA21,060 aa
Nuclease activity regionNuclease activity region · residues 81–519Helicase activity regionHelicase activity region · residues 520–1060UPF1-type SF1B helicase ATP-bindingUPF1-type SF1B helicase ATP-binding · residues 625–823UPF1-type SF1B helicase C-terminalUPF1-type SF1B helicase C-terminal · residues 824–105202004006008001,000
loading 5EAY…
PDB 5EAY ↗
shape

This is real experimental data — atoms placed by crystallography or cryo-EM, deposited by the people who did the experiment.

I have never read one of these pictures before

A protein is one long chain of amino acids that folds up. The ribbon traces that chain from one end to the other; it is not a wire, it is a simplification of where the backbone runs. Drag to rotate, scroll to zoom.

Domains are the stretches of chain that do a particular job, and they are what the coloured blocks above the picture mark out. Click one and the view flies to it. In this project the important question about any domain is usually whether the fusion still has it — the labels say lost where it does not.

Anything in yellow sticks is a bound small molecule, which is usually the drug.

Go deeper
Why this entry refuses its own headline number

A mean gene effect of -2.71 with 100% of lines below -1.0 is, on its face, the deepest dependency in this entire beat - deeper than CTCF's -1.83, which is itself a near-total pan-essential. It would be easy and wrong to write that down.

The reason it is wrong is the denominator. 80 screens in 17 tissue blocks, in the same call where seven other genes returned 1,258 in 29. Whatever selects those eighty lines is unknown to this beat, and a subset of one-sixteenth the panel, returning a distribution whose maximum (-1.55) is below every other gene's mean, does not look like a random subsample of the same panel.

The honest options are: (a) DNA2 is genuinely pan-essential and is screened in a smaller library; (b) the 80-screen row comes from a different dataset with a different scaling; (c) something else. This beat did not resolve it and does not need to, because no conclusion in this beat depends on DNA2's dependency score. What it does depend on is nobody later quoting '-2.71' beside EXO1's '-0.127' as though the two were comparable.

This is rule 83's second half working as intended: the spread in denominators is usually legitimate, and the discipline is to state the gene's own screen count every time rather than to harmonise it away.

The mitochondrial confound, which EXO1 does not have

DNA2 is annotated to both the nucleus and the mitochondrion, and neither assignment carries an evidence code. The note records a correction - originally reported as exclusively mitochondrial, later shown to be both.

That has a practical consequence for any experiment this graph might propose. A DNA2 depletion or a DNA2 inhibitor changes mitochondrial DNA maintenance as well as nuclear end resection, so a phenotype in a progenitor - which is metabolically distinctive, and which this graph's ISR and one-carbon beats have already shown is sensitive to mitochondrial state - cannot be attributed to resection without a mitochondrial control. EXO1 has no such confound: three coded nuclear citations and no other compartment.

So within the redundant EXO1/DNA2 pair, EXO1 is the cleaner experimental handle and DNA2 is the cleaner nuclease, and that asymmetry should decide which one an experiment perturbs.

Facts and figures
uniprotP51530
uniprot entry nameDNA2_HUMAN
gene symbolDNA2
gene primary readbackDNA2
protein nameDNA replication ATP-dependent helicase/nuclease DNA2
length AA1060
n PDB entries1
PDB noteOne deposit only, and no polymer entity was fetched in this beat, so nothing is claimed about what 5EAY covers, at what resolution, or whether the construct is full-length.
how the accession was resolvedgene_exact:DNA2 AND organism_id:9606 AND reviewed:true, fetched live 2026-09-22, returned EXACTLY ONE reviewed human record with gene_primary DNA2; length asserted at 1,060 aa.
localizationNucleus AND Mitochondrion, and BOTH are UNCODED - neither carries an ECO:0000269 or an ECO:0000305. The record's note reads: 'Was initially reported to be exclusively mitochondrial (PubMed:18995831). However, it was later shown to localize both in mitochondrion and nucleus (PubMed:19487465)'. This is a case where the uncoded comment carries the whole content, which is rule 91's point, and where the database is recording a historical correction rather than a settled experimental assignment.
subnuclear localizationNOT ANNOTATED. UniProt assigns DNA2 no subnuclear compartment at all. The only subnuclear statement available is functional and comes from [E2] - recruitment to laser- and nuclease-induced double-strand breaks in U2OS, downstream of CTCF-dependent early BRCA1/BARD1 and alongside SMARCAD1 and EXO1. There is no imaging of DNA2 in any haematopoietic cell, no public occupancy dataset, and nothing at the KMT2A locus in any cell type.
localization shown for fusionfalse
cell typeU2OS and HEK-293T only [E2]. 80 DepMap screens across 17 tissue blocks, of which 16 are haematopoietic or lymphoid [E3]. NO ENCODE experiment in any biosample [E5]. No KMT2A-rearranged cell, no infant material, no B-lineage progenitor, and one leukaemia record in the whole title/abstract index [E4].
cell stateS/G2 for the resection role, as for EXO1 - long-range resection commits a break to homologous recombination and requires a sister chromatid. DNA2 additionally has a replication role and a mitochondrial one, so a DNA2 perturbation changes state in three places at once, which makes it the least clean member of this cascade to manipulate and the one whose loss-of-function phenotype is hardest to attribute.
epigenetic readoutNone directly. DNA2 reads no histone mark and writes none; like EXO1 it sits in this graph's epigenetic argument purely as the EFFECTOR whose recruitment reports whether the H3K9me/H3K4me0 licence at the break was granted [E2].
shown for fusionNO - nobody has looked, in any fusion. DNA2[tiab] AND KMT2A[tiab] = 0 and DNA2[tiab] AND MLL[tiab] = 0, both with non-empty querytranslations and warninglist 'No items found.'; DNA2[tiab] AND leukemia[tiab] = 1 (PMID 24413054, which also appears in the EXO1 leukaemia set and is not about KMT2A). The bare operand DNA2[tiab] = 709.
archive bound 2026 09 22ENCODE has NO target record for DNA2 - the target.label facet returns HTTP 404 - so no released occupancy dataset exists for it in any biosample. Same for EXO1, SUV39H1 and RBBP8. CORRECTED 2026-09-22: BARD1 is NOT in this class. Measured live, /targets/BARD1-human/ returns HTTP 200 and the 404 came from pinning assay_title=TF+ChIP-seq to the query, so it described the FILTER and not the protein. RAD50 behaves identically (200 bare, 404 filtered). EXO1, DNA2, SUV39H1 and RBBP8 are genuine - 404 on both endpoints.
absence classnever-measured
domains
Nuclease activity region81.0Nuclease activity region: 81Helicase activity region520.0Helicase activity region: 520UPF1-type SF1B helicase A…625.0UPF1-type SF1B helicase ATP-binding: 625UPF1-type SF1B helicase C…824.0UPF1-type SF1B helicase C-terminal: 824
gene synonyms in uniprot
[
  "DNA2L",
  "KIAA0083"
]
PDB IDS
[
  "5EAY"
]
identifier traps
{
  "registry_check": "Rule 88 read run first over d['collisions'] (177 rows, case-sensitive): DNA2 returns zero rows. No prior trap logged.",
  "the_symbol_is_a_common_string": "'DNA2' is a four-character token containing 'DNA', and this beat did NOT title-check the 709 bare-operand records. Any future counted claim built on DNA2[tiab] must title-check, because a token of this shape in a field about DNA is exactly the recall/precision failure documented for RD (NELFE) and U19 (EAF2). The conjunctions used here return zero, so inflation does not affect this entry's conclusions - but it would affect any positive count.",
  "retired_symbol": "DNA2L was the previous symbol and was NOT searched in this beat; this is recorded as a limit on the absence, not assumed away."
}
domains — the numbers behind the chart
[
  {
    "name": "Nuclease activity region",
    "start": 81,
    "end": 519
  },
  {
    "name": "Helicase activity region",
    "start": 520,
    "end": 1060
  },
  {
    "name": "UPF1-type SF1B helicase ATP-binding",
    "start": 625,
    "end": 823
  },
  {
    "name": "UPF1-type SF1B helicase C-terminal",
    "start": 824,
    "end": 1052
  }
]
depmap
{
  "n_screens_with_gene_effect": 80,
  "n_tissue_blocks_flattened": 17,
  "mean_gene_effect": -2.7119,
  "median_gene_effect": -2.7575,
  "min_gene_effect": -3.9526,
  "max_gene_effect": -1.5485,
  "frac_below_minus_0_5": 1,
  "frac_below_minus_1_0": 1,
  "n_blood_screens": 16,
  "blood_tissue_blocks": [
    "lymphoid tissue",
    "hematopoietic tissue"
  ],
  "blood_mean_gene_effect": -2.8876,
  "blood_frac_below_minus_0_5": 1,
  "open_targets_data_version": {
    "year": "26",
    "month": "06",
    "iteration": null
  },
  "depmap_release": null,
  "THE_DENOMINATOR_ANOMALY_READ_THIS_BEFORE_USING_THE_NUMBER": "In ONE call, ONE moment and ONE Open Targets data version, DNA2 returned 80 screens across 17 tissue blocks while BRCA1, BARD1, CBX3, CBX5, KDM5A, EXO1 and CTCF each returned 1,258 across 29, and CBX1 returned 1,257. That is a sixteen-fold smaller denominator. Rule 83 states that DepMap screens each gene in a gene-specific number of lines and that most such spread is legitimate, so this is NOT being reported as a release inconsistency - but a sixteen-fold gap accompanied by a mean four times deeper than any other member of the panel is a combination that must be stated rather than ranked on. The cause was not diagnosed in this beat.",
  "rule_83": "no DepMap release identifier is present in the payload; the Open Targets data version is year 26, month 06, iteration null, and that is not a release. The gene-specific screen count is stated explicitly (80) and must never be read as a corpus denominator.",
  "rule_84": "no median-split proportion is reported, and no ranking of DNA2 against the other cascade members is made.",
  "live_zero_screen_control": "POLR2A in the same call returned 0 tissue blocks and 0 screens with a passing approvedSymbol readback, which is why an 80-screen row can be distinguished from an empty one. Without that control an unusually small row would have been indistinguishable from a partially missing one.",
  "reading": "REFUSED. DNA2's mean of -2.71 with 100% of 80 lines below -1.0 is either a genuine pan-essential signal or an artefact of whatever selects those 80 screens, and this beat cannot tell which. It is recorded, flagged, and NOT compared with EXO1's -0.127 over 1,258. See `ana-q32-depmap-across-the-hr-licensing-cascade`."
}
How this could bite us back
  • DNA2 perturbation hits nuclear end resection, DNA replication and mitochondrial DNA maintenance simultaneously, and both of its compartment assignments are uncoded in UniProt. Any phenotype attributed to its resection role without a mitochondrial control is unattributable, and a fetal haematopoietic progenitor is exactly the cell type where a mitochondrial effect would be easiest to mistake for something else.
  • Its DepMap row must not be quoted beside the other cascade members': 80 screens against 1,258 in the same call, with a panel maximum below every other gene's mean. If that number reaches a ledger it will be re-quoted as 'DNA2 is the deepest dependency in the cascade', which this beat cannot support.
  • Because EXO1 and DNA2 are redundant, neither single gene is a usable handle for testing whether the licensing cascade operates at a locus - the pair must move together, which doubles the reagent burden of any experiment this graph proposes.
Still open
  1. Why does DNA2 return 80 DepMap screens where every other gene in the same call returns 1,258? Diagnosing this is a ten-minute query against the DepMap portal directly and it decides whether a striking number is a finding or an artefact.
  2. Which of the two compartments carries DNA2's resection function in a haematopoietic progenitor, and how large is the nuclear pool? Both UniProt assignments are uncoded and the primary literature contains an explicit correction.
  3. Is DNA2 required at all in the CTCF cascade, or is the pair readout carried entirely by EXO1? The source paper never separates them.
  4. Does the retired symbol DNA2L reach a literature the current symbol misses? Not searched here.
Every source behind this entry (5)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.database

    DNA2 is UniProt P51530, entry name DNA2_HUMAN, 1,060 amino acids, primary gene name DNA2 with gene synonyms DNA2L and KIAA0083, recommended protein name 'DNA replication ATP-dependent helicase/nuclease DNA2'; the feature table annotates a UPF1-type SF1B helicase ATP-binding domain at 625-823 and its C-terminal domain at 824-1052, with regions typed for nuclease activity at 81-519 and helicase activity at 520-1060; it cross-references exactly one PDB entry, 5EAY; and its subcellular locations are Nucleus and Mitochondrion, BOTH UNCODED, with a note that it was initially reported to be exclusively mitochondrial (PubMed:18995831) and later shown to localize both in mitochondrion and nucleus (PubMed:19487465).

    What this evidence cannot show

    A database record fetched 2026-09-22 with gene_exact:DNA2 AND organism_id:9606 AND reviewed:true returning exactly one reviewed record. Both subcellular locations are UNCODED, so the database provides no experimental evidence code for either compartment - unusual, and recorded rather than smoothed over. No subnuclear compartment is annotated at all. No PDB polymer entity was fetched.

    Evidence that argues against this (1)
    • The two cited PubMed ids in the note disagree with each other by construction - one reports exclusive mitochondrial localisation and the other reports dual localisation. Neither was fetched in this beat, so which experiments support the nuclear pool, and how large that pool is, are open. A protein whose nuclear residence was contested in the primary literature is a weak foundation for a nuclear-resection argument without reading those two papers. UniProt Consortium 2026
  2. E2moderate confidence 3 of 5: One strong study, or several smaller consistent ones.paper

    DNA2 is named in the CTCF-directed HR cascade only as half of the pair EXO1/DNA2: in U2OS and HEK-293T cells, failure of the CTCF-recruited repressive-chromatin assembly abrogates recruitment of the exonucleases EXO1/DNA2, which is causally linked to defective DNA end resection and dysfunctional HR; recruitment of SMARCAD1 and EXO1/DNA2 relies on CTCF-dependent early BRCA1/BARD1 rather than on CTCF-independent late BRCA1/BARD1; and the CTCF-directed HR pathway involves both CtIP-dependent short-range resection and the subsequent EXO1/DNA2-dependent long-range resection.

    CTCF primes repressive chromatin for rapid BRCA1/BARD1 recruitment to drive homologous recombination repair of DNA double-strand breaks. · Hwang SY, Choi D, Choi MJ, Jo HR, Jeong JH, Kang MA, Lee JS · Nucleic Acids Research 2026
    PMID 42475683DOI
    What this evidence cannot show

    Confidence is set one tier BELOW the equivalent claim in gene-exo1 deliberately: the paper never separates DNA2 from EXO1, so every DNA2-specific statement here is an attribution from a pair readout. 'DNA2' appears 54 times in the full text and essentially always as 'EXO1/DNA2'. U2OS reporter lines and HEK-293T only; no haematopoietic cell, no fusion, no KMT2A locus. DOI taken from PubmedData/ArticleIdList only.

    Evidence that argues against this (1)
    • EXO1 and DNA2 are redundant for long-range resection, so a pair readout cannot establish that DNA2 contributes at all in this system; the entire effect could be EXO1's. Nothing in the paper's design separates them, and this entry should not be cited as evidence that DNA2 specifically is CTCF-dependent. Hwang SY et al. 2026
  3. E3strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    DNA2's DepMap row is measured in a sixteen-fold smaller panel than every other gene fetched in the same call, and its distribution is correspondingly incomparable: flattening every tissue block gives 80 screens across 17 tissue blocks with a gene effect, mean -2.7119, median -2.7575, minimum -3.9526, MAXIMUM -1.5485, with 100% of lines below -0.5 and 100% below -1.0; the 16 hematopoietic and lymphoid screens give a mean of -2.8876. In the same call and the same Open Targets data version, BRCA1, BARD1, CBX3, CBX5, KDM5A, EXO1 and CTCF each returned 1,258 screens across 29 tissue blocks and CBX1 returned 1,257.

    Open Targets Platform GraphQL API, target.depMapEssentiality with all tissue blocks flattened, plus meta.dataVersion · Open Targets Platform; Broad Institute DepMap · Open Targets Platform 2026
    What this evidence cannot show

    Rule 83: no DepMap release identifier is present in the payload; the Open Targets data version is year 26, month 06, iteration null. The claim asserted here is about the NUMBERS AS RETURNED, not about DNA2's biology - the cause of the 80-versus-1,258 asymmetry was not diagnosed in this beat and no ranking of DNA2 against the other cascade members is made anywhere in this entry. Rule 84: no median-split proportion is reported. The live zero-screen control fired (POLR2A: 0 blocks, 0 screens, passing symbol readback), which is what makes an 80-screen row interpretable as small rather than as broken.

    Evidence that argues against this (1)
    • The whole-panel maximum of -1.55 is below every other cascade gene's MEAN, which is not what a random 80-line subsample of the same 1,258-line panel would look like. That pattern is at least as consistent with a different underlying dataset or scaling as with a genuine pan-essentiality, and until it is diagnosed the number should not enter any ledger or comparison. Open Targets Platform; Broad Institute DepMap 2026
  4. E4strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    No published title or abstract pairs DNA2 with this graph's gene. Measured live 2026-09-22 with one [tiab] tag per term and every querytranslation read back: DNA2[tiab] AND KMT2A[tiab] returns 0 with translation '"DNA2"[Title/Abstract] AND "KMT2A"[Title/Abstract]' and warninglist outputmessages 'No items found.'; DNA2[tiab] AND MLL[tiab] returns 0 the same way; DNA2[tiab] AND leukemia[tiab] returns 1 (PMID 24413054). The bare operand DNA2[tiab] returns 709. Rule-67 positive controls through the identical SYMBOL[tiab] AND TOKEN[tiab] shape return non-zero: CTCF[tiab] AND KMT2A[tiab] 7, RAD21[tiab] AND KMT2A[tiab] 5.

    NCBI E-utilities esearch over PubMed, db=pubmed, retmode=json, querytranslation read back for every query · National Center for Biotechnology Information · NCBI Entrez Programming Utilities 2026
    What this evidence cannot show

    Titles and abstracts only - rule 80 applies. The retired symbol DNA2L was NOT searched, so the name list is incomplete. The 709 bare-operand records were not title-checked, and 'DNA2' is a short token in a field about DNA, so that figure is an upper bound on the genuine DNA2 literature; it is quoted only to establish that the operand is large, which is the rule-67 requirement when an intersection is empty.

    Evidence that argues against this (1)
    • TOP2B[tiab] AND KMT2A[tiab] also returns 0 through the identical shape although that relationship is central to this disease, because the relevant literature writes 'MLL' and 'topoisomerase II'. The DNA2 zero is therefore accurate as a measurement and cannot be promoted to 'nobody has ever looked' without a vocabulary search this beat did not run. National Center for Biotechnology Information 2026
  5. E5strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    ENCODE holds no released TF ChIP-seq experiment for DNA2 and no target record for it at all - the target.label facet returns HTTP 404 - as it does for EXO1, SUV39H1 and RBBP8. CORRECTED 2026-09-22: BARD1 is NOT in this class. Measured live, /targets/BARD1-human/ returns HTTP 200 and the 404 came from pinning assay_title=TF+ChIP-seq to the query, so it described the FILTER and not the protein. RAD50 behaves identically (200 bare, 404 filtered). EXO1, DNA2, SUV39H1 and RBBP8 are genuine - 404 on both endpoints. In the same census BRCA1 has 7, CBX3 4, CBX1 2, CBX5 4, KDM5A 3, TRIM28 5, SETDB1 5 and CTCF 502.

    What this evidence cannot show

    Counted 2026-09-22 on the released TF ChIP-seq facet only; GEO, ChIP-Atlas and ReMap were not swept. A 404 means no ENCODE target record, which is a different negative from a searched zero. As for EXO1, ChIP-seq is arguably the wrong assay class for a resection nuclease and its absence from a transcription-factor archive is only weakly informative.

    Evidence that argues against this (1)
    • Absence from a TF ChIP-seq facet is close to uninformative for a nuclease that is not a transcription factor. The meaningful archive question for DNA2 is whether any resection-mapping dataset (END-seq, BLESS, ssDNA-seq in the AsiSI/DIvA system) covers chr11q23, and that was not queried in this beat. NCBI Gene Expression Omnibus 2026
Connected to