Oncofusion

start / graph / gene-brca2

Genemoderate confidence 3 of 5: One strong study, or several smaller consistent ones.Q35-encode-reanalysis

BRCA2 (P51587, 3,418 aa) - the RAD51 loader that receives the resected end BRCA1 licensed, named in 10 entries and never written; DepMap mean -0.519, and its 12-record KMT2A intersection is entirely germline-panel and pan-cancer sequencing with no mechanism

also called FANCD1 · breast cancer type 2 susceptibility protein · BRCC2

BRCA2 is the protein that loads the machine which finds and copies an intact template after a DNA break, so it is the step that actually finishes the repair BRCA1 starts. In this leukaemia it appears only in gene-panel lists, never in an experiment.

The same thing, for a scientist

**Named in 10 entries and never written.** BRCA2 is the downstream terminus of the resection arm this graph argues about: end resection produces 3' ssDNA, and BRCA2 - via its BRC repeats and with PALB2 as the physical link to BRCA1 - displaces RPA and nucleates the RAD51 filament that performs strand invasion. **If the CTCF/HP1-gamma arm licenses BRCA1/BARD1 and BRCA1/BARD1 licenses EXO1/DNA2, BRCA2 is what the whole chain is FOR**, and it has had no entry. **The dependency.** Mean Chronos -0.519 across 1,258 lines, 53.0% below -0.5 - the second deepest in this beat's panel, behind MRE11 and ahead of BRCA1 (-0.489). **What the KMT2A intersection actually contains, title-checked.** `BRCA2[tiab] AND KMT2A[tiab]` returns **12** and `BRCA2[tiab] AND MLL[tiab]` returns **5**, both with clean translations. These are not small numbers by this graph's standards, and they are the reason this entry has to say what they are: on inspection they are germline-predisposition panels, pan-cancer genomic profiling series and ovarian/nasopharyngeal sequencing studies in which BRCA2 and KMT2A both appear as entries in a mutated-gene list. **The intersection is co-occurrence in a gene panel, not a mechanism**, and this beat found no record in it that measures BRCA2 in a KMT2A-rearranged cell or at the KMT2A locus. This is the shape rule 71 warns about in reverse: a healthy-looking count made of the wrong thing. **Not in ENCODE.** No released human BRCA2 TF ChIP-seq experiment exists, so unlike BRCA1 this one cannot be asked of the archive at all.

Facts and figures
gene symbolBRCA2
uniprotP51587
uniprot entry nameBRCA2_HUMAN
gene primary readbackBRCA2
length AA3418
protein nameBreast cancer type 2 susceptibility protein
how the accession was resolvedgene_exact:BRCA2 AND organism_id:9606 AND reviewed:true returned exactly ONE reviewed record with primary gene name BRCA2 (rules 62/65).
domains noteUniProt P51587 declares no Domain features in the fields fetched, so data.domains is EMPTY rather than populated from recall. The eight BRC repeats and the DNA-binding domain are Repeat and Region features and were not converted into domain entries here, because inventing boundaries is worse than leaving the 3D lane empty.
localizationNucleus; Chromosome; Chromosome, telomere; Cytoplasm, cytoskeleton, microtubule organizing center, centrosome (UniProt P51587)
subnuclear localizationRAD51 foci at resected double-strand breaks
localization shown for fusionNO - nobody has looked in any fusion; no ENCODE ChIP-seq experiment exists for BRCA2 in any cell type
shown for fusionNO - the 12-record KMT2A intersection is germline-panel and pan-cancer sequencing co-occurrence, with no experiment in a KMT2A-rearranged cell
cell typenot measured in a KMT2A-rearranged or infant B-lineage cell
cell statedamage-conditional
epigenetic readoutnone directly; BRCA2 acts on resected DNA, downstream of the chromatin steps this graph's cascade describes
encode statusNO released human TF ChIP-seq experiment for BRCA2 - unlike BRCA1, which has seven
absence classnever-measured
relationship to existing entriesgene-brca1 exists and covers the upstream mediator; gene-palb2, written in the same tranche, is the physical link between the two. This entry is the RAD51-loading terminus and is written so that the resection chain in this graph has a downstream end.
gene synonyms in uniprot
[
  "FACD",
  "FANCD1"
]
domains
[]
depmap
{
  "source": "Open Targets Platform 26.06 (meta.dataVersion year 26 month 06), fetched 2026-09-22",
  "gene_effect_metric": "Chronos",
  "n_screens_this_gene": 1258,
  "mean_gene_effect": -0.5186,
  "n_lines_below_minus_0_5": 667,
  "fraction_below_minus_0_5": 0.5302,
  "min": -1.4696,
  "max": 0.4805,
  "approvedSymbol_readback": "BRCA2",
  "release_note": "rule 83: no release id in the payload, so the Open Targets dataVersion is recorded. Screen count is gene-specific (1258).",
  "no_median_split": "rule 84: no anti-selective-half tally reported.",
  "live_low_screen_control": "OR4F5 (ENSG00000186092), same session: approvedSymbol reads back but only 3 tissue blocks / 5 screens are returned - a passing symbol readback is not evidence the payload is populated."
}
bounded absence 2026 09 22
{
  "instrument": "PubMed esearch, one [tiab] tag per term, every querytranslation read back",
  "BRCA2[tiab] AND KMT2A[tiab]": {
    "count": 12,
    "querytranslation": "\"BRCA2\"[Title/Abstract] AND \"KMT2A\"[Title/Abstract]",
    "character_of_the_hits": "germline predisposition panels, pan-cancer genomic profiling and ovarian/nasopharyngeal sequencing; BRCA2 and KMT2A co-occur as entries in mutated-gene lists",
    "on_topic_for_a_kmt2a_locus_or_fusion_measurement": 0
  },
  "BRCA2[tiab] AND MLL[tiab]": {
    "count": 5,
    "querytranslation": "\"BRCA2\"[Title/Abstract] AND \"MLL\"[Title/Abstract]",
    "on_topic": 0
  },
  "bounding_note": "rule 80 - this zero is bounded by TITLES AND ABSTRACTS. A BRCA2 measurement sitting in a supplementary figure of a KMT2A paper is invisible to it, and the measurements a mechanism graph wants are usually supporting panels.",
  "name_list_searched": [
    "BRCA2"
  ],
  "names_NOT_searched": [
    "FANCD1",
    "BRCC2"
  ],
  "absence_class": "never-measured"
}
Still open
  1. Is BRCA2 loss or hypomorphism enriched in KMT2A-rearranged leukaemia beyond the germline-panel co-occurrence? The 12 records are panel studies and none was designed to ask it.
  2. Does the BRCA2 intersection hide a supporting-panel measurement? Rule 80 says an abstract-bounded zero on a MEASUREMENT should be followed by opening the two or three most likely papers, and that was not done here.
  3. Would a BRCA2 or RAD51 readout discriminate between the two halves of the consecutive-resection hypothesis? If the menin arm and the CTCF arm are consecutive steps of one reaction, a downstream RAD51-focus readout should be equally sensitive to both perturbations, which is a cheaper test than the ssDNA-resection qPCR the hypothesis currently proposes.
Every source behind this entry (2)
  1. E1established confidence 5 of 5: Multiple independent lines of evidence; treated as settled in the field.paper

    Purified human BRCA2 stimulates RAD51-mediated homologous recombination, establishing directly that BRCA2's function is to load and activate the RAD51 recombinase on resected single-stranded DNA rather than to act earlier in the pathway.

    PMID 20729832DOI
    What this evidence cannot show

    A purified-protein biochemical reconstitution. It establishes the activity, not its regulation in a cell, and says nothing about chromatin context, cell type or this locus.

    Evidence that argues against this (1)
    • A reconstitution with purified components cannot report on what limits the reaction in vivo, where BRCA2 loading depends on PALB2, BRCA1 and the resection that produced the substrate - so the step this graph is actually arguing about is upstream of everything this paper measures. Zhang F et al. 2009
  2. E2strong confidence 4 of 5: Replicated in peer-reviewed work by more than one group.database

    BRCA2 has a mean DepMap Chronos gene effect of -0.519 across 1,258 lines with 53.0% below -0.5, placing it second only to MRE11 among the resection-cascade genes measured in this beat.

    DepMap CRISPR gene-effect (Chronos) for BRCA2, served by the Open Targets Platform GraphQL API · Broad Institute DepMap / Open Targets Platform · Open Targets Platform 26.06 (meta.dataVersion year 26 month 06), fetched 2026-09-22 2026
    What this evidence cannot show

    Rule 83: Open Targets dataVersion 26.06 recorded in place of a DepMap release id; gene-specific screen count. Rule 84: no median-split tally. Cultured-line fitness, not a safety or tractability claim, and BRCA2-mutant tumours are the clinical setting where loss is the therapeutic opportunity rather than a liability.

    Evidence that argues against this (1)
    • A deep mean across all lineages is the opposite of a selectivity signal, and this graph's own DepMap guidance warns that a gene depleted in most lines has little dynamic range left in which KMT2A-specific selectivity could appear. BRCA2 at 53% of lines below -0.5 is approaching that floor-compression regime. Broad Institute DepMap / Open Targets Platform 2026
Connected to