# C. elegans domain profile
# Filename (without .yaml) is the domain key: "c-elegans"
# Sources: WormBook nomenclature, CGC strain naming conventions

name: "C. elegans"

journals:
  - Genetics
  - "PLOS Genetics"
  - Development
  - "Cell Reports"
  - "Nature"
  - eLife

# Species
species:
  first_mention: "Caenorhabditis elegans"
  subsequent: ["the worm", "C. elegans", "worms", "nematodes"]
  avoid: "the nematode Caenorhabditis elegans"

# Stocks — CGC strains
stocks:
  format: "N2"
  rrid_prefix: "RRID:WB-STRAIN_"
  description: "Caenorhabditis Genetics Center (CGC)"
  rules:
    - "Strain names: 2-3 uppercase letters + number (e.g. N2, CB112, BA17)"
    - "Lab code identifies the originating lab"
    - "Wild type is always N2 (Bristol strain)"

# Genotype format
genotype:
  format: "BA17 (fem-1(hc17_ts)) IV"
  rules:
    - "Roman numerals indicate linkage group (I, II, III, IV, V, X)"
    - "Genes in parentheses within genotype"
    - "Mutation alleles in parentheses"

# Nomenclature — Brenner 1974, Horvitz 1979
nomenclature:
  - rule: "Gene names: 3 italic lowercase letters + hyphen + number (e.g. _let-37_, _dpy-5_, _unc-54_)"
  - rule: "Optional roman numeral suffix for linkage group (e.g. _dpy-5 I_)"
  - rule: "Protein names: 3-letter uppercase or specific protein name, NOT italic (e.g. LET-37)"
  - rule: "Allele names: 2 letters + number (e.g. hc17)"
  - rule: "Temperature-sensitive alleles: suffix _ts (e.g. hc17_ts)"
  - rule: "Wild type allele: + sign"
  - rule: "Transgenes: NOT italicized (e.g. Ex[dpy-5p::GFP])"
  - rule: "Strain names: 2-3 uppercase lab code + number (e.g. N2, CB112, MT2542)"

# Balancers
balancers:
  canonical: [hT2, nT1, sT1, mIn1, eT1, qC1]
  rules:
    - "C. elegans balancers are translocation chromosomes"
    - "Use format: +/hT2 or +/nT1 for heterozygotes"

# Auto-detect keywords
detect_keywords: ["C\\. elegans", Caenorhabditis, worm, nematode, "let-", "dpy-", "unc-", "egl-", N2, CGC, WormBase, WormBook]

# Term mappings
term_mappings:
  - source: "nematode"
    target: "C. elegans"
  - source: "worm model"
    target: "C. elegans"

# Standard assays
standard_assays:
  - "lethality screen"
  - "RNAi feeding"
  - "microparticle bombardment"
  - "CRISPR/Cas9 editing"
  - "MosSCI insertion"
  - "aging assay"
  - "locomotion assay"
  - "heat stress assay"

# Life stages
life_stages:
  - embryo
  - "L1 larva"
  - "L2 larva"
  - "L3 larva"
  - "L4 larva"
  - adult
  - dauer
  - "aging day 1 adult"
  - "day 1 post-L4"

# Reporting standards
reporting:
  rrid_required: true
  rrid_types: [antibody, strain, "plasmid"]
  key_resources_table: true
  acknowledge_cgc: "Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440)."
  acknowledge_wormbase: "Gene information from WormBase (wormbase.org)."

# Voice rules
voice:
  introduction: "HIGH assertiveness. State the gap; state the question."
  methods: "HIGHEST assertiveness. Detail strain, allele, temperature."
  results: "HIGH assertiveness. State findings directly."
  discussion: "Moderate hedging. Speculation only in final paragraph."

# Sex
sex:
  - hermaphrodite
  - male