/**
 * Canonical oncology biomarkers — explicitly curated metadata (CLAUDE.md §8, SPEC §17, §52, §121).
 *
 * What this file IS: the identity of each biomarker entity (slug, name, kind, anchor gene, NCIt
 * concept, aliases) and a neutral description of what is measured and how the result is used
 * clinically (diagnostic / predictive / prognostic). Assay families and scoring conventions are
 * recorded as metadata with links to the authority pages they come from.
 *
 * What this file is NOT: it carries no clinical-utility grading, no thresholds of its own, no
 * dosing and no individual advice. Every link biomarker → cancer / drug / trial / approval /
 * publication is DERIVED AT QUERY TIME from source-native records (CIViC evidence, knowledge
 * edges, ClinicalTrials.gov, openFDA / Health Canada, PubMed) — see docs/methodology/biomarkers.md.
 *
 * NCIt codes: EVERY code below was verified live against the NCI EVS REST API on 2026-09-11
 * (`GET https://api-evsrest.nci.nih.gov/api/v1/concept/ncit/<code>?include=minimal`), NCIt version
 * 26.08e; each returned `active: true`, `conceptStatus: DEFAULT`, and the name recorded in
 * `ncitName` is the preferred name returned by the API. Where NCIt has a biomarker / gene-alteration
 * concept ("… Positive", "… Gene Mutation", "… Fusion Positive", "p.V600E"…) that concept is used
 * (`ncitConceptKind: 'biomarker'`). No row falls back to the bare gene concept.
 *
 * Gene anchors: `geneSymbol` is the HGNC symbol resolved to `genes.id` at seed time (seed.ts reports
 * unresolved symbols and never invents). `genes` lists additional HGNC symbols for multi-gene
 * markers (MSI-H/dMMR → MMR genes, NTRK → NTRK1/2/3, BCR-ABL1 → BCR + ABL1). `variantSlugs` anchors
 * molecular-level markers to variant entities present in `variants` (checked on 2026-09-11 against
 * the production copy); when present, derived links are restricted to those variants instead of
 * the whole gene.
 *
 * Tumor-agnostic flag: `tumorAgnostic: true` is a curated statement that an FDA tissue-agnostic
 * indication exists for this marker. The UI/API never assert it alone — they list the actual
 * `drug_approvals` rows with `tumor_agnostic = true` reached through the derived drug set (or an
 * indication-text match on `indicationTerms`), and say when none is ingested yet.
 */

export const BIOMARKER_KINDS = ['gene_mutation', 'protein_expression', 'hormone_receptor', 'immune_marker', 'msi', 'tmb', 'hrd', 'ctdna', 'methylation', 'signature', 'cell_surface', 'other'] as const;
export type BiomarkerKind = (typeof BIOMARKER_KINDS)[number];

export const BIOMARKER_KIND_LABEL: Record<BiomarkerKind, string> = {
  gene_mutation: 'Gene alteration',
  protein_expression: 'Protein expression',
  hormone_receptor: 'Hormone receptor',
  immune_marker: 'Immune marker',
  msi: 'Microsatellite instability / MMR',
  tmb: 'Tumor mutational burden',
  hrd: 'Homologous recombination deficiency',
  ctdna: 'Circulating tumor DNA',
  methylation: 'Methylation',
  signature: 'Signature',
  cell_surface: 'Cell-surface target',
  other: 'Other',
};

/** Verification record stored on every row (`measurement.verification`) and shown in the UI. */
export const NCIT_VERIFICATION = {
  authority: 'NCI Thesaurus (NCIt) via the EVS REST API',
  endpoint: 'https://api-evsrest.nci.nih.gov/api/v1/concept/ncit/{code}?include=minimal',
  ncitVersion: '26.08e',
  verifiedAt: '2026-09-11',
} as const;

/** Authority pages cited in `measurement.sources` (all returned HTTP 200 on 2026-09-11). */
export const BIOMARKER_SOURCES = {
  fdaCdx: { label: 'FDA — List of cleared or approved companion diagnostic devices', url: 'https://www.fda.gov/medical-devices/in-vitro-diagnostics/list-cleared-or-approved-companion-diagnostic-devices-in-vitro-and-imaging-tools' },
  nciBiomarkerTesting: { label: 'NCI — Biomarker testing for cancer treatment', url: 'https://www.cancer.gov/about-cancer/treatment/types/biomarker-testing-cancer-treatment' },
  nciTumorMarkers: { label: 'NCI — Tumor markers fact sheet', url: 'https://www.cancer.gov/about-cancer/diagnosis-staging/diagnosis/tumor-markers-fact-sheet' },
  nciCheckpoint: { label: 'NCI — Immune checkpoint inhibitors', url: 'https://www.cancer.gov/about-cancer/treatment/types/immunotherapy/checkpoint-inhibitors' },
  fdaOncologyApprovals: { label: 'FDA — Oncology (cancer) / hematologic malignancies approval notifications', url: 'https://www.fda.gov/drugs/resources-information-approved-drugs/oncology-cancer-hematologic-malignancies-approval-notifications' },
} as const;
type SourceKey = keyof typeof BIOMARKER_SOURCES;

export type AssayFamily = 'IHC' | 'FISH' | 'ISH' | 'NGS' | 'PCR' | 'RT-PCR' | 'Sequencing' | 'Flow cytometry' | 'Karyotype' | 'PET imaging' | 'Liquid biopsy' | 'MSI-PCR' | 'Fragment analysis';

export interface BiomarkerSeed {
  slug: string;
  name: string;
  kind: BiomarkerKind;
  /** HGNC symbol of the anchor gene, resolved to genes.id at seed time; null for non-gene markers. */
  geneSymbol: string | null;
  /** Additional HGNC symbols the marker depends on (multi-gene markers). */
  genes?: string[];
  /** Variant slugs present in `variants` that define a molecular-level marker. */
  variantSlugs?: string[];
  ncitCode: string;
  ncitName: string;
  ncitConceptKind: 'biomarker';
  aliases: string[];
  /** One neutral sentence: what is measured and how it is used (diagnostic / predictive / prognostic). */
  description: string;
  assays: AssayFamily[];
  /** Scoring / reporting conventions as published by the authority (no thresholds of our own). */
  scoring?: string;
  notes?: string;
  sources: SourceKey[];
  /** Curated statement that an FDA tissue-agnostic indication exists (verified against approval rows at query time). */
  tumorAgnostic?: boolean;
  /** Terms matched (case-insensitive, whole phrase) against approval indication text at query time. */
  indicationTerms?: string[];
}

const S = BIOMARKER_SOURCES;

export const BIOMARKER_SEED: BiomarkerSeed[] = [
  // ── Breast / hormone / HER2 ─────────────────────────────────────────────────────────────────
  {
    slug: 'her2',
    name: 'HER2 (ERBB2) overexpression / amplification',
    kind: 'protein_expression',
    geneSymbol: 'ERBB2',
    variantSlugs: ['erbb2-amplification', 'erbb2-overexpression', 'erbb2-overexpression-civic-875'],
    ncitCode: 'C68748',
    ncitName: 'HER2/Neu Positive',
    ncitConceptKind: 'biomarker',
    aliases: ['HER2', 'HER2/neu', 'ERBB2', 'HER2-positive', 'HER2 amplification', 'HER2-low'],
    description: 'HER2 protein overexpression or ERBB2 gene amplification measured on tumor tissue; used as a predictive marker for HER2-directed therapy in breast, gastric/gastro-esophageal and other carcinomas.',
    assays: ['IHC', 'ISH', 'FISH', 'NGS'],
    scoring: 'IHC scored 0, 1+, 2+ or 3+ (2+ is equivocal and reflexed to in situ hybridization); ISH reported as HER2/CEP17 ratio and mean HER2 copy number; "HER2-low" denotes IHC 1+ or IHC 2+/ISH-negative in the breast label wording.',
    sources: ['fdaCdx', 'nciBiomarkerTesting'],
    tumorAgnostic: true,
    indicationTerms: ['HER2-positive', 'HER2 (IHC 3+)', 'HER2-low', 'HER2-overexpressing', 'HER2 overexpressing', 'HER2-expressing'],
  },
  {
    slug: 'er',
    name: 'Estrogen receptor (ER)',
    kind: 'hormone_receptor',
    geneSymbol: 'ESR1',
    ncitCode: 'C15492',
    ncitName: 'Estrogen Receptor Positive',
    ncitConceptKind: 'biomarker',
    aliases: ['ER', 'ER-positive', 'ESR1 expression', 'estrogen receptor alpha'],
    description: 'Estrogen receptor protein expression measured by immunohistochemistry on tumor tissue; used as a predictive marker for endocrine therapy and as part of breast cancer subtyping.',
    assays: ['IHC'],
    scoring: 'Reported as percentage of nuclei staining and intensity (Allred / H-score conventions); "hormone receptor-positive" in labels denotes ER- and/or PR-positive disease.',
    sources: ['nciBiomarkerTesting', 'nciTumorMarkers'],
    indicationTerms: ['hormone receptor (HR)-positive', 'hormone receptor-positive', 'estrogen receptor-positive', 'ER-positive'],
  },
  {
    slug: 'pr',
    name: 'Progesterone receptor (PR)',
    kind: 'hormone_receptor',
    geneSymbol: 'PGR',
    ncitCode: 'C15496',
    ncitName: 'Progesterone Receptor Positive',
    ncitConceptKind: 'biomarker',
    aliases: ['PR', 'PgR', 'PR-positive', 'PGR expression'],
    description: 'Progesterone receptor protein expression measured by immunohistochemistry on tumor tissue; used together with ER as a predictive and prognostic marker in breast cancer.',
    assays: ['IHC'],
    scoring: 'Reported as percentage of nuclei staining and intensity, alongside ER.',
    sources: ['nciBiomarkerTesting', 'nciTumorMarkers'],
    indicationTerms: ['progesterone receptor-positive', 'PR-positive'],
  },
  {
    slug: 'erbb2-mutation',
    name: 'HER2 (ERBB2) activating mutation',
    kind: 'gene_mutation',
    geneSymbol: 'ERBB2',
    variantSlugs: ['erbb2-mutation', 'erbb2-exon-20-insertion'],
    ncitCode: 'C96866',
    ncitName: 'ERBB2 Gene Mutation',
    ncitConceptKind: 'biomarker',
    aliases: ['HER2 mutation', 'ERBB2 mutation', 'HER2 exon 20 insertion', 'HER2-mutant NSCLC'],
    description: 'Activating ERBB2 (HER2) mutations, chiefly exon 20 insertions, detected by sequencing of tumor tissue or plasma; used as a predictive marker distinct from HER2 overexpression, notably in non-small cell lung cancer.',
    assays: ['NGS', 'Liquid biopsy'],
    notes: 'Distinct from HER2 protein overexpression / gene amplification (see biomarker "her2").',
    sources: ['fdaCdx', 'nciBiomarkerTesting'],
    indicationTerms: ['HER2 (ERBB2)-mutant', 'HER2 (ERBB2) activating mutations', 'HER2-mutant'],
  },
  {
    slug: 'esr1-mutation',
    name: 'ESR1 mutation',
    kind: 'gene_mutation',
    geneSymbol: 'ESR1',
    variantSlugs: ['esr1-mutation', 'esr1-d538g', 'esr1-y537s'],
    ncitCode: 'C136629',
    ncitName: 'ESR1 Gene Mutation',
    ncitConceptKind: 'biomarker',
    aliases: ['ESR1-mutated', 'ESR1 ligand-binding domain mutation', 'D538G', 'Y537S'],
    description: 'Acquired mutations in the ESR1 ligand-binding domain detected in tumor tissue or circulating tumor DNA; used as a predictive marker in ER-positive breast cancer after endocrine therapy.',
    assays: ['NGS', 'PCR', 'Liquid biopsy'],
    sources: ['fdaCdx', 'nciBiomarkerTesting'],
    indicationTerms: ['ESR1-mutated', 'ESR1 mutation'],
  },
  {
    slug: 'pik3ca-mutation',
    name: 'PIK3CA mutation',
    kind: 'gene_mutation',
    geneSymbol: 'PIK3CA',
    variantSlugs: ['pik3ca-mutation', 'pik3ca-h1047r', 'pik3ca-e545k'],
    ncitCode: 'C96271',
    ncitName: 'PIK3CA Gene Mutation',
    ncitConceptKind: 'biomarker',
    aliases: ['PIK3CA-mutated', 'PIK3CA-altered', 'H1047R', 'E545K'],
    description: 'Activating PIK3CA mutations detected in tumor tissue or plasma; used as a predictive marker for PI3K-pathway inhibitors, notably in hormone receptor-positive breast cancer.',
    assays: ['NGS', 'PCR', 'Liquid biopsy'],
    sources: ['fdaCdx', 'nciBiomarkerTesting'],
    indicationTerms: ['PIK3CA-mutated', 'PIK3CA mutation'],
  },
  {
    slug: 'ki-67',
    name: 'Ki-67 labeling index',
    kind: 'protein_expression',
    geneSymbol: 'MKI67',
    ncitCode: 'C157250',
    ncitName: 'Ki-67 Labeling Index',
    ncitConceptKind: 'biomarker',
    aliases: ['Ki-67', 'Ki67', 'MKI67', 'proliferation index'],
    description: 'Percentage of tumor nuclei staining for the proliferation antigen Ki-67 by immunohistochemistry; used as a prognostic marker and, in some breast cancer indications, as an eligibility criterion.',
    assays: ['IHC'],
    scoring: 'Reported as percentage of positive nuclei; laboratory-specific cut-offs, not standardized across indications.',
    sources: ['nciTumorMarkers', 'fdaCdx'],
    indicationTerms: ['Ki-67'],
  },

  // ── Lung and other solid tumors: driver alterations ─────────────────────────────────────────
  {
    slug: 'egfr-mutation',
    name: 'EGFR mutation (any sensitizing or resistance mutation)',
    kind: 'gene_mutation',
    geneSymbol: 'EGFR',
    ncitCode: 'C98357',
    ncitName: 'EGFR Gene Mutation',
    ncitConceptKind: 'biomarker',
    aliases: ['EGFR-mutant', 'EGFR-mutated', 'EGFR activating mutation'],
    description: 'Somatic mutations of the EGFR tyrosine kinase domain detected by sequencing or PCR in tumor tissue or plasma; used as a predictive marker for EGFR tyrosine kinase inhibitors, chiefly in non-small cell lung cancer.',
    assays: ['NGS', 'PCR', 'Liquid biopsy'],
    notes: 'Gene-level entry; exon 19 deletion, L858R, T790M and exon 20 insertion are separate molecular markers below.',
    sources: ['fdaCdx', 'nciBiomarkerTesting'],
    indicationTerms: ['EGFR exon 19 deletions or exon 21 L858R', 'EGFR mutation', 'EGFR-mutated', 'EGFR mutations'],
  },
  { slug: 'egfr-exon-19-deletion', name: 'EGFR exon 19 deletion', kind: 'gene_mutation', geneSymbol: 'EGFR', variantSlugs: ['egfr-exon-19-deletion'], ncitCode: 'C126892', ncitName: 'EGFR Exon 19 Deletion Mutation', ncitConceptKind: 'biomarker', aliases: ['ex19del', 'EGFR del19', 'exon 19 del'], description: 'In-frame deletions in EGFR exon 19 detected by sequencing or PCR; a sensitizing EGFR mutation used as a predictive marker for EGFR tyrosine kinase inhibitors in non-small cell lung cancer.', assays: ['NGS', 'PCR', 'Liquid biopsy'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['exon 19 deletions'] },
  { slug: 'egfr-l858r', name: 'EGFR L858R', kind: 'gene_mutation', geneSymbol: 'EGFR', variantSlugs: ['egfr-l858r'], ncitCode: 'C98515', ncitName: 'EGFR NP_005219.2:p.L858R', ncitConceptKind: 'biomarker', aliases: ['p.L858R', 'exon 21 L858R', 'c.2573T>G'], description: 'EGFR exon 21 point mutation p.L858R detected by sequencing or PCR; a sensitizing EGFR mutation used as a predictive marker for EGFR tyrosine kinase inhibitors in non-small cell lung cancer.', assays: ['NGS', 'PCR', 'Liquid biopsy'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['L858R'] },
  { slug: 'egfr-t790m', name: 'EGFR T790M', kind: 'gene_mutation', geneSymbol: 'EGFR', variantSlugs: ['egfr-t790m'], ncitCode: 'C98503', ncitName: 'EGFR NP_005219.2:p.T790M', ncitConceptKind: 'biomarker', aliases: ['p.T790M', 'gatekeeper mutation', 'c.2369C>T'], description: 'EGFR exon 20 point mutation p.T790M detected in tumor tissue or plasma; an acquired resistance mutation to earlier-generation EGFR inhibitors used as a predictive marker for third-generation agents.', assays: ['NGS', 'PCR', 'Liquid biopsy'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['T790M'] },
  { slug: 'egfr-exon-20-insertion', name: 'EGFR exon 20 insertion', kind: 'gene_mutation', geneSymbol: 'EGFR', variantSlugs: ['egfr-exon-20-insertion'], ncitCode: 'C125605', ncitName: 'EGFR Exon 20 Insertion Mutation', ncitConceptKind: 'biomarker', aliases: ['ex20ins', 'EGFR exon 20 ins'], description: 'In-frame insertions in EGFR exon 20 detected by sequencing; a class of EGFR mutations with distinct sensitivity used as a predictive marker in non-small cell lung cancer.', assays: ['NGS', 'Liquid biopsy'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['exon 20 insertion'] },
  {
    slug: 'alk-fusion',
    name: 'ALK rearrangement / fusion',
    kind: 'gene_mutation',
    geneSymbol: 'ALK',
    ncitCode: 'C142135',
    ncitName: 'ALK Fusion Positive',
    ncitConceptKind: 'biomarker',
    aliases: ['ALK-positive', 'ALK rearrangement', 'EML4-ALK', 'ALK fusion'],
    description: 'ALK gene rearrangements (most often EML4-ALK) detected by FISH, immunohistochemistry or sequencing; used as a predictive marker for ALK inhibitors in non-small cell lung cancer and some lymphomas.',
    assays: ['FISH', 'IHC', 'NGS', 'RT-PCR'],
    notes: 'Gene-level entry: derived links cover every ALK fusion partner recorded as a variant.',
    sources: ['fdaCdx', 'nciBiomarkerTesting'],
    indicationTerms: ['anaplastic lymphoma kinase (ALK)-positive', 'ALK-positive', 'ALK rearrangement'],
  },
  { slug: 'ros1-fusion', name: 'ROS1 rearrangement / fusion', kind: 'gene_mutation', geneSymbol: 'ROS1', ncitCode: 'C131071', ncitName: 'ROS1 Fusion Positive', ncitConceptKind: 'biomarker', aliases: ['ROS1-positive', 'ROS1 rearrangement', 'CD74-ROS1'], description: 'ROS1 gene rearrangements detected by FISH, immunohistochemistry or sequencing; used as a predictive marker for ROS1-directed tyrosine kinase inhibitors in non-small cell lung cancer.', assays: ['FISH', 'IHC', 'NGS', 'RT-PCR'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['ROS1-positive', 'ROS1 rearrangement'] },
  { slug: 'kras-g12c', name: 'KRAS G12C', kind: 'gene_mutation', geneSymbol: 'KRAS', variantSlugs: ['kras-g12c'], ncitCode: 'C98365', ncitName: 'KRAS NP_004976.2:p.G12C', ncitConceptKind: 'biomarker', aliases: ['p.G12C', 'KRAS G12C-mutated', 'c.34G>T'], description: 'KRAS codon 12 point mutation p.G12C detected by sequencing or PCR in tumor tissue or plasma; used as a predictive marker for KRAS G12C inhibitors in non-small cell lung and colorectal cancer.', assays: ['NGS', 'PCR', 'Liquid biopsy'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['KRAS G12C'] },
  { slug: 'kras-mutation', name: 'KRAS mutation (any)', kind: 'gene_mutation', geneSymbol: 'KRAS', ncitCode: 'C41361', ncitName: 'KRAS Gene Mutation', ncitConceptKind: 'biomarker', aliases: ['KRAS-mutant', 'KRAS-mutated', 'RAS mutation'], description: 'Activating KRAS mutations (codons 12, 13, 61 and others) detected by sequencing or PCR; used as a predictive marker, including as a negative predictor of response to anti-EGFR antibodies in colorectal cancer.', assays: ['NGS', 'PCR', 'Liquid biopsy'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['KRAS wild-type', 'RAS wild-type', 'KRAS-mutated', 'KRAS mutation'] },
  { slug: 'nras-mutation', name: 'NRAS mutation', kind: 'gene_mutation', geneSymbol: 'NRAS', ncitCode: 'C41381', ncitName: 'NRAS Gene Mutation', ncitConceptKind: 'biomarker', aliases: ['NRAS-mutant', 'NRAS-mutated'], description: 'Activating NRAS mutations detected by sequencing or PCR; used as a predictive marker in colorectal cancer (anti-EGFR antibodies) and studied in melanoma.', assays: ['NGS', 'PCR'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['NRAS wild-type', 'NRAS mutation'] },
  {
    slug: 'braf-v600e',
    name: 'BRAF V600E',
    kind: 'gene_mutation',
    geneSymbol: 'BRAF',
    variantSlugs: ['braf-v600e'],
    ncitCode: 'C98342',
    ncitName: 'BRAF NP_004324.2:p.V600E',
    ncitConceptKind: 'biomarker',
    aliases: ['p.V600E', 'BRAF V600E-mutant', 'c.1799T>A'],
    description: 'BRAF codon 600 point mutation p.V600E detected by sequencing, PCR or mutation-specific immunohistochemistry; used as a predictive marker for BRAF/MEK inhibitor therapy across melanoma, thyroid, colorectal, lung and other cancers.',
    assays: ['NGS', 'PCR', 'IHC', 'Liquid biopsy'],
    sources: ['fdaCdx', 'nciBiomarkerTesting'],
    tumorAgnostic: true,
    indicationTerms: ['BRAF V600E', 'BRAF V600'],
  },
  { slug: 'braf-mutation', name: 'BRAF mutation (any)', kind: 'gene_mutation', geneSymbol: 'BRAF', ncitCode: 'C40430', ncitName: 'BRAF Gene Mutation', ncitConceptKind: 'biomarker', aliases: ['BRAF-mutant', 'BRAF-mutated', 'non-V600 BRAF', 'class II/III BRAF'], description: 'Any BRAF mutation (V600 and non-V600 classes) detected by sequencing; used as a predictive and, in some settings, prognostic marker.', assays: ['NGS', 'PCR'], notes: 'Gene-level entry; V600E is a separate molecular marker.', sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['BRAF mutation', 'BRAF-mutated', 'BRAF wild-type'] },
  { slug: 'met-exon-14-skipping', name: 'MET exon 14 skipping', kind: 'gene_mutation', geneSymbol: 'MET', variantSlugs: ['met-exon-14-skipping-mutation', 'met-exon-14-mutation'], ncitCode: 'C131179', ncitName: 'MET Exon 14 Skipping Mutation', ncitConceptKind: 'biomarker', aliases: ['METex14', 'MET exon 14 alteration', 'METΔ14'], description: 'MET alterations causing exon 14 skipping detected by RNA- or DNA-based sequencing; used as a predictive marker for MET tyrosine kinase inhibitors in non-small cell lung cancer.', assays: ['NGS', 'RT-PCR'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['MET exon 14 skipping'] },
  { slug: 'met-amplification', name: 'MET amplification', kind: 'gene_mutation', geneSymbol: 'MET', variantSlugs: ['met-amplification'], ncitCode: 'C43532', ncitName: 'c-MET Gene Amplification', ncitConceptKind: 'biomarker', aliases: ['MET amp', 'c-MET amplification', 'MET copy-number gain'], description: 'Increased MET gene copy number detected by FISH or sequencing; studied as a predictive marker and as a resistance mechanism to EGFR inhibitors in non-small cell lung cancer.', assays: ['FISH', 'NGS'], scoring: 'Reported as MET/CEP7 ratio or gene copy number; thresholds are assay- and study-specific.', sources: ['nciBiomarkerTesting'], indicationTerms: ['MET amplification'] },
  {
    slug: 'ret-fusion',
    name: 'RET fusion',
    kind: 'gene_mutation',
    geneSymbol: 'RET',
    ncitCode: 'C131069',
    ncitName: 'RET Fusion Positive',
    ncitConceptKind: 'biomarker',
    aliases: ['RET rearrangement', 'RET fusion-positive', 'KIF5B-RET', 'CCDC6-RET'],
    description: 'RET gene fusions detected by sequencing or FISH; used as a predictive marker for selective RET inhibitors in non-small cell lung, thyroid and other solid tumors.',
    assays: ['NGS', 'FISH', 'RT-PCR'],
    sources: ['fdaCdx', 'nciBiomarkerTesting'],
    tumorAgnostic: true,
    indicationTerms: ['RET gene fusion', 'RET fusion-positive', 'RET-mutant'],
  },
  {
    slug: 'ntrk-fusion',
    name: 'NTRK gene fusion',
    kind: 'gene_mutation',
    geneSymbol: null,
    genes: ['NTRK1', 'NTRK2', 'NTRK3'],
    ncitCode: 'C183255',
    ncitName: 'NTRK Gene Fusion Positive',
    ncitConceptKind: 'biomarker',
    aliases: ['NTRK fusion-positive', 'TRK fusion', 'ETV6-NTRK3', 'NTRK1/2/3 fusion'],
    description: 'Fusions involving NTRK1, NTRK2 or NTRK3 detected by sequencing, FISH or pan-TRK immunohistochemistry; used as a tissue-agnostic predictive marker for TRK inhibitors in solid tumors.',
    assays: ['NGS', 'FISH', 'IHC', 'RT-PCR'],
    notes: 'No single anchor gene: derived links use the three NTRK genes listed.',
    sources: ['fdaCdx', 'nciBiomarkerTesting'],
    tumorAgnostic: true,
    indicationTerms: ['NTRK', 'neurotrophic tyrosine receptor kinase', 'neurotrophic receptor tyrosine kinase'],
  },
  { slug: 'fgfr2-fusion', name: 'FGFR2 fusion / rearrangement', kind: 'gene_mutation', geneSymbol: 'FGFR2', variantSlugs: ['fgfr2-fusion', 'fgfr2-v-fusion'], ncitCode: 'C150616', ncitName: 'FGFR2 Fusion Positive', ncitConceptKind: 'biomarker', aliases: ['FGFR2 rearrangement', 'FGFR2-BICC1', 'FGFR2 fusion-positive'], description: 'FGFR2 gene fusions or rearrangements detected by sequencing or FISH; used as a predictive marker for FGFR inhibitors in cholangiocarcinoma.', assays: ['NGS', 'FISH', 'RT-PCR'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['FGFR2 fusion', 'FGFR2 rearrangement'] },
  { slug: 'fgfr3-alteration', name: 'FGFR3 alteration (mutation or fusion)', kind: 'gene_mutation', geneSymbol: 'FGFR3', ncitCode: 'C150618', ncitName: 'FGFR3 Gene Alteration Positive', ncitConceptKind: 'biomarker', aliases: ['FGFR3 mutation', 'FGFR3 fusion', 'FGFR3-TACC3', 'S249C'], description: 'Activating FGFR3 point mutations or fusions detected by sequencing of tumor tissue; used as a predictive marker for FGFR inhibitors in urothelial carcinoma.', assays: ['NGS', 'RT-PCR'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['FGFR3', 'FGFR3 genetic alterations'] },

  // ── Immune / genomic-instability markers ────────────────────────────────────────────────────
  {
    slug: 'pd-l1',
    name: 'PD-L1 expression',
    kind: 'immune_marker',
    geneSymbol: 'CD274',
    variantSlugs: ['cd274-expression', 'cd274-overexpression'],
    ncitCode: 'C128554',
    ncitName: 'PD-L1 Positive',
    ncitConceptKind: 'biomarker',
    aliases: ['PD-L1', 'CD274', 'B7-H1', 'programmed death-ligand 1', 'PD-L1 TPS', 'PD-L1 CPS'],
    description: 'Programmed death-ligand 1 protein expression measured by immunohistochemistry on tumor and/or immune cells; used as a predictive marker for PD-1/PD-L1 checkpoint inhibitors in many carcinomas.',
    assays: ['IHC'],
    scoring: 'Reported as Tumor Proportion Score (TPS, % tumor cells), Combined Positive Score (CPS, PD-L1-positive tumor and immune cells per 100 tumor cells) or immune-cell (IC) percentage; scores and cut-offs are assay- and indication-specific as stated in each label.',
    notes: 'Gene CD274 encodes PD-L1; derived links use CIViC evidence recorded on CD274 expression.',
    sources: ['fdaCdx', 'nciCheckpoint', 'nciBiomarkerTesting'],
    indicationTerms: ['PD-L1'],
  },
  {
    slug: 'msi-h',
    name: 'Microsatellite instability-high (MSI-H)',
    kind: 'msi',
    geneSymbol: null,
    genes: ['MLH1', 'MSH2', 'MSH6', 'PMS2'],
    ncitCode: 'C36493',
    ncitName: 'High-Frequency Microsatellite Instability',
    ncitConceptKind: 'biomarker',
    aliases: ['MSI-H', 'MSI-high', 'microsatellite instability', 'MSI'],
    description: 'Length instability of microsatellite loci detected by PCR or sequencing of tumor DNA, the genomic consequence of mismatch-repair deficiency; used as a tissue-agnostic predictive marker for immune checkpoint inhibitors and as a screening marker for Lynch syndrome.',
    assays: ['MSI-PCR', 'NGS', 'Liquid biopsy'],
    scoring: 'PCR panels report MSI-H, MSI-L or MSS from the number of unstable loci; NGS reports an MSI score from panel-wide microsatellite sites.',
    notes: 'MSI-H and dMMR are two measurements of one biological state: MSI-H is the DNA phenotype (PCR/NGS), dMMR is loss of MMR protein expression (IHC of MLH1, MSH2, MSH6, PMS2). Labels often say "MSI-H or dMMR".',
    sources: ['fdaCdx', 'nciCheckpoint', 'nciBiomarkerTesting'],
    tumorAgnostic: true,
    indicationTerms: ['microsatellite instability-high', 'MSI-H'],
  },
  {
    slug: 'dmmr',
    name: 'Mismatch repair deficiency (dMMR)',
    kind: 'msi',
    geneSymbol: null,
    genes: ['MLH1', 'MSH2', 'MSH6', 'PMS2'],
    ncitCode: 'C136712',
    ncitName: 'Mismatch Repair Deficiency',
    ncitConceptKind: 'biomarker',
    aliases: ['dMMR', 'MMR-deficient', 'MMR deficiency', 'loss of MMR protein expression'],
    description: 'Loss of nuclear expression of one or more mismatch-repair proteins (MLH1, MSH2, MSH6, PMS2) by immunohistochemistry; used as a tissue-agnostic predictive marker for immune checkpoint inhibitors and as a screening marker for Lynch syndrome.',
    assays: ['IHC'],
    scoring: 'Reported per protein as retained or lost nuclear staining with internal control; MLH1 loss is commonly followed by MLH1 promoter methylation / BRAF V600E testing to separate sporadic from hereditary cases.',
    notes: 'See biomarker "msi-h" for the relationship between dMMR (protein, IHC) and MSI-H (DNA phenotype, PCR/NGS).',
    sources: ['fdaCdx', 'nciCheckpoint', 'nciBiomarkerTesting'],
    tumorAgnostic: true,
    indicationTerms: ['mismatch repair deficient', 'dMMR'],
  },
  {
    slug: 'tmb-h',
    name: 'Tumor mutational burden-high (TMB-H)',
    kind: 'tmb',
    geneSymbol: null,
    ncitCode: 'C156025',
    ncitName: 'High Tumor Mutation Burden',
    ncitConceptKind: 'biomarker',
    aliases: ['TMB-H', 'TMB-high', 'tumor mutation burden', 'TMB'],
    description: 'Number of somatic mutations per megabase of sequenced tumor genome estimated from a large sequencing panel; used as a tissue-agnostic predictive marker for PD-1 blockade in solid tumors.',
    assays: ['NGS', 'Liquid biopsy'],
    scoring: 'Reported in mutations per megabase (mut/Mb); the FDA-approved tissue-agnostic indication uses the cut-off of the companion diagnostic named in the label; values are not comparable across panels.',
    notes: 'A genome-wide measurement with no anchor gene: gene-derived blocks are empty by construction; approvals are matched on indication text.',
    sources: ['fdaCdx', 'nciCheckpoint', 'nciBiomarkerTesting'],
    tumorAgnostic: true,
    indicationTerms: ['tumor mutational burden-high', 'tumor mutational burden', 'TMB-H'],
  },
  {
    slug: 'hrd',
    name: 'Homologous recombination deficiency (HRD)',
    kind: 'hrd',
    geneSymbol: null,
    ncitCode: 'C120465',
    ncitName: 'Homologous Recombination Deficiency',
    ncitConceptKind: 'biomarker',
    aliases: ['HRD', 'HRD-positive', 'genomic instability score', 'HRR deficiency'],
    description: 'A genomic scar signature (loss of heterozygosity, telomeric allelic imbalance, large-scale state transitions) and/or deleterious BRCA1/2 status estimated from tumor sequencing; used as a predictive marker for PARP inhibitors in ovarian cancer.',
    assays: ['NGS'],
    scoring: 'Reported as a genomic instability score with an assay-specific positivity threshold, combined with BRCA1/2 mutation status; not comparable across assays.',
    notes: 'A genome-wide signature with no single anchor gene (BRCA1/2 are separate biomarkers): gene-derived blocks are empty by construction; approvals are matched on indication text.',
    sources: ['fdaCdx', 'nciBiomarkerTesting'],
    indicationTerms: ['homologous recombination deficiency', 'HRD-positive', 'HRD positive'],
  },
  {
    slug: 'ctdna',
    name: 'Circulating tumor DNA (ctDNA)',
    kind: 'ctdna',
    geneSymbol: null,
    ncitCode: 'C113243',
    ncitName: 'Circulating Tumor-Derived DNA',
    ncitConceptKind: 'biomarker',
    aliases: ['ctDNA', 'liquid biopsy', 'cell-free tumor DNA', 'plasma genotyping'],
    description: 'Tumor-derived DNA fragments in plasma analyzed by sequencing or PCR; used as a specimen for genotyping (mutations, fusions, MSI, TMB) and studied as a prognostic marker for minimal residual disease.',
    assays: ['Liquid biopsy', 'NGS', 'PCR'],
    notes: 'A specimen type rather than a single analyte: derived blocks have no gene to anchor to; molecular markers detected in plasma appear under their own entries.',
    sources: ['nciBiomarkerTesting', 'fdaCdx'],
    indicationTerms: ['plasma specimen', 'circulating tumor DNA'],
  },

  // ── Hereditary / DNA-repair genes ───────────────────────────────────────────────────────────
  { slug: 'brca1-mutation', name: 'BRCA1 mutation', kind: 'gene_mutation', geneSymbol: 'BRCA1', ncitCode: 'C19635', ncitName: 'BRCA1 Gene Mutation', ncitConceptKind: 'biomarker', aliases: ['BRCA1-mutated', 'gBRCA1', 'germline BRCA1', 'somatic BRCA1'], description: 'Pathogenic germline or somatic BRCA1 variants detected by sequencing of blood or tumor; used as a predictive marker for PARP inhibitors and as a hereditary cancer-risk marker.', assays: ['NGS', 'Sequencing'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['BRCA-mutated', 'BRCA1', 'gBRCAm', 'BRCA mutation'] },
  { slug: 'brca2-mutation', name: 'BRCA2 mutation', kind: 'gene_mutation', geneSymbol: 'BRCA2', ncitCode: 'C19636', ncitName: 'BRCA2 Gene Mutation', ncitConceptKind: 'biomarker', aliases: ['BRCA2-mutated', 'gBRCA2', 'germline BRCA2', 'somatic BRCA2'], description: 'Pathogenic germline or somatic BRCA2 variants detected by sequencing of blood or tumor; used as a predictive marker for PARP inhibitors and as a hereditary cancer-risk marker.', assays: ['NGS', 'Sequencing'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['BRCA-mutated', 'BRCA2', 'gBRCAm', 'BRCA mutation'] },
  { slug: 'tp53-mutation', name: 'TP53 mutation', kind: 'gene_mutation', geneSymbol: 'TP53', ncitCode: 'C118396', ncitName: 'TP53 Gene Mutation', ncitConceptKind: 'biomarker', aliases: ['TP53-mutated', 'p53 mutation', 'TP53-mutant'], description: 'Somatic TP53 mutations detected by sequencing of tumor tissue or marrow; used mainly as a prognostic and risk-stratification marker across solid and hematologic malignancies.', assays: ['NGS', 'Sequencing', 'IHC'], sources: ['nciBiomarkerTesting'], indicationTerms: ['TP53 mutation'] },
  { slug: 'cdkn2a-deletion', name: 'CDKN2A deletion / loss', kind: 'gene_mutation', geneSymbol: 'CDKN2A', variantSlugs: ['cdkn2a-deletion', 'cdkn2a-loss'], ncitCode: 'C41615', ncitName: 'CDKN2A Gene Deletion', ncitConceptKind: 'biomarker', aliases: ['CDKN2A loss', 'p16 loss', 'CDKN2A homozygous deletion', '9p21 deletion'], description: 'Homozygous deletion or loss of CDKN2A (p16INK4a) detected by FISH, sequencing or p16 immunohistochemistry; used as a diagnostic and prognostic marker in mesothelioma, glioma and other tumors.', assays: ['FISH', 'NGS', 'IHC'], sources: ['nciBiomarkerTesting'] },

  // ── Prostate / neuroendocrine cell-surface targets ──────────────────────────────────────────
  { slug: 'ar', name: 'Androgen receptor (AR)', kind: 'hormone_receptor', geneSymbol: 'AR', ncitCode: 'C94297', ncitName: 'Androgen Receptor Positive', ncitConceptKind: 'biomarker', aliases: ['AR', 'AR-positive', 'androgen receptor expression'], description: 'Androgen receptor expression or alteration assessed by immunohistochemistry or sequencing; the therapeutic target of androgen-deprivation and AR-pathway inhibitors in prostate cancer and a subtyping marker in breast cancer.', assays: ['IHC', 'NGS'], sources: ['nciBiomarkerTesting', 'nciTumorMarkers'] },
  { slug: 'ar-v7', name: 'AR-V7 splice variant', kind: 'other', geneSymbol: 'AR', variantSlugs: ['ar-ar-v7'], ncitCode: 'C135616', ncitName: 'Androgen Receptor Splice Variant 7 Positive', ncitConceptKind: 'biomarker', aliases: ['AR-V7', 'androgen receptor variant 7', 'AR splice variant 7'], description: 'A constitutively active androgen receptor splice variant lacking the ligand-binding domain, detected in circulating tumor cells or tissue by RT-PCR or immunostaining; studied as a predictive marker of resistance to AR-pathway inhibitors in castration-resistant prostate cancer.', assays: ['RT-PCR', 'IHC', 'Liquid biopsy'], sources: ['nciBiomarkerTesting'] },
  { slug: 'psma', name: 'PSMA (FOLH1) expression', kind: 'cell_surface', geneSymbol: 'FOLH1', ncitCode: 'C153464', ncitName: 'FOLH1 Positive', ncitConceptKind: 'biomarker', aliases: ['PSMA', 'FOLH1', 'prostate-specific membrane antigen', 'PSMA-positive', 'GCPII'], description: 'Prostate-specific membrane antigen expression on tumor cells, assessed in vivo by PSMA-ligand PET imaging; used as an eligibility marker for PSMA-targeted radioligand therapy in metastatic castration-resistant prostate cancer.', assays: ['PET imaging', 'IHC'], scoring: 'PET positivity defined relative to liver uptake and the absence of PSMA-negative lesions, per the imaging agent label.', sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['PSMA-positive', 'prostate-specific membrane antigen (PSMA)-positive'] },
  { slug: 'sstr', name: 'Somatostatin receptor (SSTR) expression', kind: 'cell_surface', geneSymbol: 'SSTR2', ncitCode: 'C128873', ncitName: 'Somatostatin Receptor Positive', ncitConceptKind: 'biomarker', aliases: ['SSTR', 'SSTR2', 'somatostatin receptor-positive', 'SSTR imaging'], description: 'Somatostatin receptor (chiefly SSTR2) expression on neuroendocrine tumor cells assessed by somatostatin-analog PET/SPECT imaging or immunohistochemistry; used as an eligibility marker for somatostatin analogs and peptide receptor radionuclide therapy.', assays: ['PET imaging', 'IHC'], notes: 'SSTR2 is the anchor gene; the receptor family has five members and imaging agents bind several of them.', sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['somatostatin receptor-positive', 'somatostatin receptor positive'] },

  // ── Hematologic cell-surface targets ────────────────────────────────────────────────────────
  { slug: 'cd19', name: 'CD19', kind: 'cell_surface', geneSymbol: 'CD19', ncitCode: 'C129255', ncitName: 'CD19 Positive', ncitConceptKind: 'biomarker', aliases: ['CD19-positive', 'CD19 antigen', 'B-lymphocyte antigen CD19'], description: 'B-lineage surface antigen detected by flow cytometry or immunohistochemistry; the target antigen of CD19-directed antibodies, bispecifics and CAR T-cell therapies in B-cell leukemias and lymphomas.', assays: ['Flow cytometry', 'IHC'], sources: ['nciBiomarkerTesting', 'fdaOncologyApprovals'], indicationTerms: ['CD19-positive', 'CD19-directed'] },
  { slug: 'cd20', name: 'CD20 (MS4A1)', kind: 'cell_surface', geneSymbol: 'MS4A1', ncitCode: 'C128631', ncitName: 'CD20 Positive', ncitConceptKind: 'biomarker', aliases: ['CD20', 'CD20-positive', 'MS4A1', 'B-lymphocyte antigen CD20'], description: 'B-lineage surface antigen detected by flow cytometry or immunohistochemistry; the target antigen of anti-CD20 antibodies and CD20xCD3 bispecifics in B-cell lymphomas and chronic lymphocytic leukemia.', assays: ['Flow cytometry', 'IHC'], sources: ['nciBiomarkerTesting', 'fdaOncologyApprovals'], indicationTerms: ['CD20-positive', 'CD20-directed'] },
  { slug: 'bcma', name: 'BCMA (TNFRSF17)', kind: 'cell_surface', geneSymbol: 'TNFRSF17', ncitCode: 'C128845', ncitName: 'TNFRSF17 Positive', ncitConceptKind: 'biomarker', aliases: ['BCMA', 'B-cell maturation antigen', 'TNFRSF17', 'CD269'], description: 'Plasma-cell surface antigen detected by flow cytometry or immunohistochemistry; the target antigen of BCMA-directed antibody-drug conjugates, bispecifics and CAR T-cell therapies in multiple myeloma.', assays: ['Flow cytometry', 'IHC'], sources: ['nciBiomarkerTesting', 'fdaOncologyApprovals'], indicationTerms: ['BCMA', 'B-cell maturation antigen'] },
  { slug: 'cd38', name: 'CD38', kind: 'cell_surface', geneSymbol: 'CD38', ncitCode: 'C147087', ncitName: 'CD38 Positive', ncitConceptKind: 'biomarker', aliases: ['CD38-positive', 'ADP-ribosyl cyclase 1'], description: 'Surface glycoprotein highly expressed on plasma cells, detected by flow cytometry; the target antigen of anti-CD38 antibodies in multiple myeloma.', assays: ['Flow cytometry', 'IHC'], sources: ['nciBiomarkerTesting', 'fdaOncologyApprovals'], indicationTerms: ['CD38'] },
  { slug: 'cd30', name: 'CD30 (TNFRSF8)', kind: 'cell_surface', geneSymbol: 'TNFRSF8', ncitCode: 'C129257', ncitName: 'TNFRSF8 Positive', ncitConceptKind: 'biomarker', aliases: ['CD30', 'CD30-positive', 'TNFRSF8', 'Ki-1 antigen'], description: 'Activation antigen expressed on Hodgkin/Reed-Sternberg and anaplastic large-cell lymphoma cells, detected by immunohistochemistry; the target antigen of CD30-directed antibody-drug conjugates and a diagnostic marker.', assays: ['IHC', 'Flow cytometry'], sources: ['nciBiomarkerTesting', 'fdaOncologyApprovals'], indicationTerms: ['CD30-expressing', 'CD30-positive', 'CD30-directed'] },
  { slug: 'cd33', name: 'CD33', kind: 'cell_surface', geneSymbol: 'CD33', ncitCode: 'C132228', ncitName: 'CD33 Positive', ncitConceptKind: 'biomarker', aliases: ['CD33-positive', 'Siglec-3', 'myeloid cell surface antigen CD33'], description: 'Myeloid surface antigen detected by flow cytometry on leukemic blasts; the target antigen of CD33-directed antibody-drug conjugates in acute myeloid leukemia.', assays: ['Flow cytometry'], sources: ['nciBiomarkerTesting', 'fdaOncologyApprovals'], indicationTerms: ['CD33-positive'] },

  // ── Hematologic molecular markers ───────────────────────────────────────────────────────────
  { slug: 'flt3-mutation', name: 'FLT3 mutation (ITD / TKD)', kind: 'gene_mutation', geneSymbol: 'FLT3', ncitCode: 'C128919', ncitName: 'FLT3 Gene Mutation', ncitConceptKind: 'biomarker', aliases: ['FLT3-ITD', 'FLT3-TKD', 'FLT3 internal tandem duplication', 'FLT3-mutated', 'D835'], description: 'FLT3 internal tandem duplications or tyrosine kinase domain point mutations detected by PCR fragment analysis or sequencing of blood or marrow; used as a predictive marker for FLT3 inhibitors and a prognostic marker in acute myeloid leukemia.', assays: ['PCR', 'Fragment analysis', 'NGS'], scoring: 'ITD reported with allelic ratio (mutant/wild-type signal) where the assay supports it; TKD reported by codon (D835, I836).', notes: 'Gene-level entry covering ITD and TKD variants recorded in `variants` (e.g. slug flt3-itd).', sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['FLT3 mutation-positive', 'FLT3 mutation', 'FLT3-ITD'] },
  { slug: 'idh1-mutation', name: 'IDH1 mutation', kind: 'gene_mutation', geneSymbol: 'IDH1', ncitCode: 'C118389', ncitName: 'IDH1 Gene Mutation', ncitConceptKind: 'biomarker', aliases: ['IDH1-mutated', 'IDH1 R132', 'R132H', 'R132C'], description: 'IDH1 codon 132 mutations detected by PCR or sequencing (or R132H-specific immunohistochemistry in glioma); used as a predictive marker for IDH1 inhibitors in acute myeloid leukemia and cholangiocarcinoma and as a diagnostic marker in glioma.', assays: ['PCR', 'NGS', 'IHC'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['IDH1 mutation', 'IDH1-mutated', 'IDH1 or IDH2 mutation'] },
  { slug: 'idh2-mutation', name: 'IDH2 mutation', kind: 'gene_mutation', geneSymbol: 'IDH2', ncitCode: 'C118390', ncitName: 'IDH2 Gene Mutation', ncitConceptKind: 'biomarker', aliases: ['IDH2-mutated', 'R140Q', 'R172K'], description: 'IDH2 codon 140 or 172 mutations detected by PCR or sequencing of blood or marrow; used as a predictive marker for IDH2 inhibitors in acute myeloid leukemia and as a diagnostic marker in glioma.', assays: ['PCR', 'NGS'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['IDH2 mutation', 'IDH2-mutated', 'IDH1 or IDH2 mutation'] },
  { slug: 'npm1-mutation', name: 'NPM1 mutation', kind: 'gene_mutation', geneSymbol: 'NPM1', variantSlugs: ['npm1-mutation', 'npm1-exon-11-mutation'], ncitCode: 'C82429', ncitName: 'NPM1 Gene Mutation', ncitConceptKind: 'biomarker', aliases: ['NPM1-mutated', 'NPMc+', 'NPM1 exon 12 mutation'], description: 'Frameshift insertions in the last exon of NPM1 detected by PCR or sequencing of blood or marrow; used as a diagnostic, prognostic and measurable-residual-disease marker in acute myeloid leukemia.', assays: ['PCR', 'NGS', 'Fragment analysis'], sources: ['nciBiomarkerTesting'], indicationTerms: ['NPM1 mutation', 'NPM1-mutated'] },
  { slug: 'kit-mutation', name: 'KIT mutation', kind: 'gene_mutation', geneSymbol: 'KIT', ncitCode: 'C39712', ncitName: 'KIT Gene Mutation', ncitConceptKind: 'biomarker', aliases: ['c-KIT mutation', 'KIT exon 11', 'KIT D816V', 'CD117'], description: 'Activating KIT mutations (exon 9, 11, 13, 17 and others) detected by sequencing of tumor tissue or marrow; used as a predictive marker for KIT inhibitors in gastrointestinal stromal tumor and as a diagnostic marker in systemic mastocytosis.', assays: ['NGS', 'PCR', 'IHC'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['Kit (CD117) positive', 'KIT mutation', 'KIT D816V'] },
  { slug: 'pdgfra-mutation', name: 'PDGFRA mutation', kind: 'gene_mutation', geneSymbol: 'PDGFRA', variantSlugs: ['pdgfra-d842v'], ncitCode: 'C39718', ncitName: 'PDGFRA Gene Mutation', ncitConceptKind: 'biomarker', aliases: ['PDGFRA D842V', 'PDGFRA exon 18', 'PDGFRA-mutant'], description: 'Activating PDGFRA mutations, notably exon 18 D842V, detected by sequencing of tumor tissue; used as a predictive marker for PDGFRA-selective inhibitors in gastrointestinal stromal tumor.', assays: ['NGS', 'PCR'], sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['PDGFRA exon 18', 'PDGFRA D842V', 'PDGFRA mutation'] },
  { slug: 'bcr-abl1', name: 'BCR-ABL1 fusion (Philadelphia chromosome)', kind: 'gene_mutation', geneSymbol: 'ABL1', genes: ['BCR'], variantSlugs: ['bcr-abl1-fusion'], ncitCode: 'C94600', ncitName: 'BCR/ABL1 Fusion Gene', ncitConceptKind: 'biomarker', aliases: ['BCR-ABL1', 'BCR::ABL1', 'Philadelphia chromosome', 'Ph+', 't(9;22)', 'p210', 'p190'], description: 'The BCR-ABL1 fusion from t(9;22) detected by karyotype, FISH or quantitative RT-PCR of blood or marrow; the defining diagnostic marker of chronic myeloid leukemia, a predictive marker for ABL tyrosine kinase inhibitors, and a molecular-response marker.', assays: ['RT-PCR', 'FISH', 'Karyotype', 'NGS'], scoring: 'Quantitative RT-PCR reported on the International Scale (BCR-ABL1 %IS); kinase-domain mutations (e.g. T315I) reported separately.', sources: ['fdaCdx', 'nciBiomarkerTesting'], indicationTerms: ['Philadelphia chromosome-positive', 'Ph+', 'BCR-ABL', 'T315I'] },
  { slug: 'jak2-v617f', name: 'JAK2 V617F', kind: 'gene_mutation', geneSymbol: 'JAK2', variantSlugs: ['jak2-v617f'], ncitCode: 'C105908', ncitName: 'JAK2 NP_004963.1:p.V617F', ncitConceptKind: 'biomarker', aliases: ['p.V617F', 'JAK2-positive', 'c.1849G>T'], description: 'JAK2 exon 14 point mutation p.V617F detected by allele-specific PCR or sequencing of blood; a diagnostic marker of myeloproliferative neoplasms (polycythemia vera, essential thrombocythemia, primary myelofibrosis).', assays: ['PCR', 'NGS'], scoring: 'Reported qualitatively or as variant allele fraction.', sources: ['nciBiomarkerTesting'], indicationTerms: ['JAK2'] },
  { slug: 'del-17p', name: '17p deletion (del(17p))', kind: 'signature', geneSymbol: null, genes: ['TP53'], variantSlugs: ['tp53-deletion', 'tp53-loss'], ncitCode: 'C36499', ncitName: 'Loss of Chromosome 17p', ncitConceptKind: 'biomarker', aliases: ['del(17p)', 'del17p', '17p13 deletion', 'TP53 deletion'], description: 'Loss of the short arm of chromosome 17 (including the TP53 locus) detected by FISH or karyotype of blood or marrow; a prognostic and treatment-selection marker in chronic lymphocytic leukemia and multiple myeloma.', assays: ['FISH', 'Karyotype', 'NGS'], notes: 'A cytogenetic marker: TP53 is the gene it is anchored to for derived links; TP53 point mutations are the separate biomarker "tp53-mutation".', sources: ['nciBiomarkerTesting'], indicationTerms: ['17p deletion', 'del(17p)'] },
];

/** The measurement jsonb stored on each row — the seed is the only writer of this shape. */
export interface BiomarkerMeasurement {
  assays: AssayFamily[];
  scoring?: string;
  notes?: string;
  sources: Array<{ label: string; url: string }>;
  aliases: string[];
  genes: string[];
  variantSlugs: string[];
  indicationTerms: string[];
  tumorAgnostic: boolean;
  ncit: { code: string; name: string; conceptKind: 'biomarker' };
  verification: typeof NCIT_VERIFICATION;
}

export function buildMeasurement(b: BiomarkerSeed): BiomarkerMeasurement {
  const m: BiomarkerMeasurement = {
    assays: b.assays,
    sources: b.sources.map((k) => ({ label: BIOMARKER_SOURCES[k].label, url: BIOMARKER_SOURCES[k].url })),
    aliases: b.aliases,
    genes: b.genes ?? [],
    variantSlugs: b.variantSlugs ?? [],
    indicationTerms: b.indicationTerms ?? [],
    tumorAgnostic: b.tumorAgnostic === true,
    ncit: { code: b.ncitCode, name: b.ncitName, conceptKind: b.ncitConceptKind },
    verification: NCIT_VERIFICATION,
  };
  if (b.scoring) m.scoring = b.scoring;
  if (b.notes) m.notes = b.notes;
  return m;
}

/** Every HGNC symbol the seed needs resolved (anchor genes + multi-gene lists), unique. */
export function seedGeneSymbols(seed: BiomarkerSeed[] = BIOMARKER_SEED): string[] {
  const out = new Set<string>();
  for (const b of seed) {
    if (b.geneSymbol) out.add(b.geneSymbol);
    for (const g of b.genes ?? []) out.add(g);
  }
  return [...out].sort();
}