HLA-B - the class I heavy chain the bivalency screen was actually read out on, and the locus that presents the one KMT2A epitope this graph has a structure for

[
  "HLAB"
]

{
  "pmid": "36635503",
  "antibodies": [
    "W6/32 (pan HLA-A,B,C)",
    "REA143 (HLA class I Bw6, described as HLA-B-specific)"
  ],
  "cell_line": "K-562",
  "note": "Bw6 is a public serological epitope, not a locus-specific genetic readout",
  "title": "Targeting Menin disrupts the KMT2A/B and polycomb balance to paradoxically activate bivalent genes.",
  "authors": "Sparbier CE, Gillespie A, Gomez J, et al.",
  "venue": "Nat Cell Biol",
  "year": 2023,
  "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}

{
  "allele": "HLA-B*07:02",
  "peptide": "EPR(pS)PSHSM",
  "kmt2a_residues": "747-755",
  "phospho_site": "KMT2A S750",
  "tcr": "TCR27",
  "pmid": "37353482",
  "graph_entry": "interface-hla-b0702-kmt2a-phosphopeptide-747-755",
  "title": "Molecular mechanism of phosphopeptide neoantigen immunogenicity.",
  "authors": "Patskovsky Y, Natarajan A, Patskovska L, et al.",
  "venue": "Nat Commun",
  "year": 2023,
  "title_provenance": "Title/authors/venue backfilled from PubMed esummary on 2026-09-20 because this citation stored a bare PMID with no title, which made it unscoreable by verify-citations.py. BACKFILLING IS NOT VERIFICATION: it confirms the PMID resolves to a real paper, not that this is the right paper for the claim. That still requires reading."
}

{
  "token": "TAP2",
  "class": "alias_of_two_different_genes",
  "candidates": [
    {
      "accession": "Q03519",
      "symbol": "TAP2",
      "length_aa": 686
    },
    {
      "accession": "Q9UDX4",
      "symbol": "SEC14L3",
      "length_aa": 400,
      "why": "Tocopherol-Associated Protein 2"
    }
  ],
  "length_defence_usable": true,
  "verified": "2026-09-20"
}

{
  "chronos_mean_kmt2a_rearranged": -0.1271,
  "chronos_mean_other_leukaemia": -0.0784,
  "difference": -0.0487,
  "pan_essentiality_all_lines": -0.0601,
  "selectivity_rank": 3584,
  "rank_denominator": 17728,
  "provenance": "Joined 2026-09-21 from analyses/depmap-kmt2a-selective-scan-full (generated 2026-09-19T10:12:06Z), the project's own genome-wide scan over 11 KMT2A-rearranged and 43 other leukaemia DepMap lines. NOT re-fetched and NOT independent of that scan: every caveat it carries applies here, including that the KMT2A-rearranged arm is a small number of lines drawn from nine individuals, that the rank denominator is 17,728 genes, and that a difference is not a dependency. Added because only 26 of 308 gene entries carried this number, which made any KMT2A-selectivity question answerable for 8% of the corpus.",
  "how_to_read": "A negative difference means MORE required in KMT2A-rearranged lines than in other leukaemia lines. A rank near 17,728 is ANTI-selective. Neither is evidence of a fusion-driven dependency on its own: check pan_essentiality_all_lines before calling anything selective, and note that a paralogue-redundant gene can score near zero for reasons that have nothing to do with this disease - see ana-the-paralogue-redundancy-excuse-is-not-tested-and-my-proxy-decided-the-answer."
}

{
  "subcellular": "PLASMA MEMBRANE, single-pass type I, with an endoplasmic-reticulum-membrane pool during assembly. UniProt P01889 (fetched live 2026-09-21) carries 'Cell membrane' on THREE experimental citations (ECO:0000269 PubMed:25480565, 26439010, 9620674) with topology 'Single-pass type I membrane protein' (ECO:0000255), and 'Endoplasmic reticulum membrane' at ECO:0000305 PubMed:9620674 with the same topology. The feature table settles the topology independently: SIGNAL PEPTIDE 1-24 and a HELICAL TRANSMEMBRANE SEGMENT 310-333 of a 362-residue protein [E_q93_hlab_loc1].",
  "subnuclear": "NOT APPLICABLE. HLA-B is a type I membrane glycoprotein of the secretory pathway with a cleaved N-terminal signal peptide and a single C-terminal transmembrane helix; it is co-translationally inserted into the ER membrane, assembles with beta-2-microglobulin and peptide, and is displayed at the plasma membrane. There is no soluble nuclear species, so there is no subnuclear address to determine. Recorded as a CATEGORY EXCLUSION so it stays OUT of the denominator of proteins whose subnuclear compartment nobody has measured - the handling Q87 gave B2M and PROM1, Q89 gave CD33 and ELOVL1, and earlier tranches gave HLA-A, NRAS and SEPTIN2. THE DISPATCH BRIEF PREDICTED THIS OUTCOME FOR HLA-B AND THE PREDICTION IS CONFIRMED.",
  "not_applicable_reason": "secretory-pathway type I membrane glycoprotein, signal peptide 1-24 and transmembrane 310-333, no nuclear pool",
  "AND_ONE_THING_THE_PREDICTION_DID_NOT_CONTAIN": "NONE OF THE THREE EXPERIMENTAL CITATIONS FOR THE CELL-MEMBRANE VALUE IS A LOCALISATION STUDY. PubMed:25480565 is about KIR3DL1*001 interaction depending on molecular microarchitecture within the Bw4 epitope, 26439010 is TAPBPR functioning as a peptide-exchange catalyst, and 9620674 is HLA-B27-restricted antigen presentation in the absence of tapasin [E_q93_hlab_loc2]. All three are peptide-loading and receptor-recognition immunology; a surface location is presupposed throughout rather than measured. For THIS protein it does not matter, because the topology annotation settles the compartment independently of any citation - and that is the useful general point: rule 91's audit is worth running even where the answer cannot change, because it identifies WHICH annotations are load-bearing. Here the signal peptide and the transmembrane helix are, and the citations are not.",
  "how_shown": "topology from the sequence features (signal peptide 1-24, transmembrane 310-333); the surface location is presupposed by the three cited immunology studies rather than measured in them; ER-membrane assembly pool at ECO:0000305, curator inference",
  "the_readout_is_the_identifier": "Brief rule 74 governs every surface claim on this gene and this entry already carries it: REA143 is anti-Bw6, a serological specificity on a SUBSET of HLA-B alleles with Bw4 additionally carried by some HLA-A alleles, and W6/32 detects the ASSEMBLED heavy-chain/beta-2-microglobulin heterodimer and is therefore silently a B2M-dependent, assembly-and-trafficking readout. Note that P01889's own cell-membrane citations include a Bw4-epitope microarchitecture study and a TAPBPR peptide-exchange study, so even the database's evidence for 'this protein is at the surface' is allele- and assembly-scoped.",
  "trafficking_is_the_regulated_step": "As for HLA-A, the interesting localisation question is not where the protein rests but HOW MUCH completes the ER-to-surface journey. Loss of B2M, TAP1/TAP2 or tapasin retains heavy chain in the ER, and PubMed:9620674 is precisely a study of what happens to HLA-B27 presentation WITHOUT TAPASIN - so this gene's cited evidence is about the trafficking step rather than the address.",
  "node": "localization-plasma-membrane-leukaemic-surface-antigens"
}

{
  "haematopoietic_answer": "ALL NUCLEATED HAEMATOPOIETIC CELLS, and therefore on leukaemic blasts, as one of the three classical class I heavy chains. HLA-B is the most polymorphic of them (this entry's existing E1), so the informative unit is the ALLELE rather than the cell type - which serological epitopes (Bw4/Bw6) and which KIR ligands a given patient presents is a germline property.",
  "in_this_disease": "This entry's existing content is the cell-type record and it is specific: HLA-B is the gene the bivalency screen was read out on, used as the browser and CUT&Tag exemplar of a bivalent MHC-I gene, and it is the locus presenting the one KMT2A epitope this graph has a structure for. Both are in KMT2A-rearranged material.",
  "why_a_cell_type_field_is_the_wrong_instrument_here": "Between haematopoietic cell types the classical class I heavy chains are flat, exactly as gene-hla-a records. What varies is the allele and the SURFACE AMOUNT in a given leukaemic state - under interferon, or under a menin inhibitor - and neither is a cell-type property.",
  "cross_reference": "read with gene-hla-a (the paralogue, three residues apart in length and undetectable by the length check), gene-b2m (the invariant light chain, and the reason a W6/32 readout is assembly-dependent) and gene-nlrc5 (the transactivator arm)."
}

{
  "state": "INTERFERON-RESPONSIVE AND MENIN-INHIBITOR-RESPONSIVE SURFACE ABUNDANCE, not a resting state - the same framing gene-hla-a carries, and see cell-state-interferon-herv-mhc2 for the interferon arm.",
  "the_state_that_is_specific_to_THIS_gene": "BIVALENT CHROMATIN. This entry's existing content records HLA-B as the exemplar of a bivalent MHC-I gene in the screen the whole beat rests on, which makes the relevant state cell-state-primed-bivalent-chromatin: the gene is held poised rather than off, and the surface protein amount is downstream of a chromatin state this project studies directly.",
  "and_the_state_variable_is_ASSEMBLY_not_only_TRANSCRIPTION": "Tapasin-independent and TAPBPR-mediated peptide exchange decide how much assembled heterodimer reaches the surface [E_q93_hlab_loc2], so a transcript-level state measurement and a W6/32 flow measurement can move apart. Brief rule 74's point, restated as a cell-state caution."
}

{
  "date": "2026-09-21",
  "tool": "NCBI E-utilities esearch, db=pubmed",
  "run_by": "Q93",
  "queries": [
    {
      "query": "\"HLA-B\"[tiab]",
      "querytranslation": "\"HLA-B\"[Title/Abstract]",
      "count": 8580,
      "role": "operand control"
    },
    {
      "query": "HLA-B[tiab] AND (twelve-term subnuclear clause)",
      "count": 1,
      "title_check": "PMID 41697750, immune correlates of HIV-1 rebound during broadly neutralising antibody treatment in young children - off topic.",
      "bound_not_reproducible_as_stored": "NOT REPRODUCIBLE AS STORED, flagged 2026-09-23. This `query` field records a SUMMARY of the search rather than the search: the clause text it names was never written down, so the string cannot be re-run and this bound could not be included in the cycle-71 audit that re-ran 40 of the corpus's other zero-bounds against the live index (all 40 still returned zero). The clauses have deliberately NOT been reconstructed - guessing them would manufacture a query that was never run, which is worse than an abbreviated one because it would look auditable. This flag does not dispute the count or the absence; it records that this particular bound is not independently checkable."
    },
    {
      "query": "HLA-B[tiab] AND (nine-term KMT2A/MLL fusion clause)",
      "count": 1,
      "title_check": "PMID 42144646.",
      "bound_not_reproducible_as_stored": "NOT REPRODUCIBLE AS STORED, flagged 2026-09-23. This `query` field records a SUMMARY of the search rather than the search: the clause text it names was never written down, so the string cannot be re-run and this bound could not be included in the cycle-71 audit that re-ran 40 of the corpus's other zero-bounds against the live index (all 40 still returned zero). The clauses have deliberately NOT been reconstructed - guessing them would manufacture a query that was never run, which is worse than an abbreviated one because it would look auditable. This flag does not dispute the count or the absence; it records that this particular bound is not independently checkable."
    },
    {
      "query": "HLA-B[tiab] AND (localisation clause) AND (fusion clause)",
      "count": 0,
      "role": "reported for completeness only; the zero bounds a question that is a category error for a type I membrane protein",
      "bound_not_reproducible_as_stored": "NOT REPRODUCIBLE AS STORED, flagged 2026-09-23. This `query` field records a SUMMARY of the search rather than the search: the clause text it names was never written down, so the string cannot be re-run and this bound could not be included in the cycle-71 audit that re-ran 40 of the corpus's other zero-bounds against the live index (all 40 still returned zero). The clauses have deliberately NOT been reconstructed - guessing them would manufacture a query that was never run, which is worse than an abbreviated one because it would look auditable. This flag does not dispute the count or the absence; it records that this particular bound is not independently checkable."
    }
  ],
  "reading": "Titles and abstracts only (brief rule 80). The counts are recorded so the NOT APPLICABLE call is auditable rather than asserted: 8,580 records on this gene and ONE subnuclear hit, which is off topic.",
  "absence_class": "not-applicable"
}