biotender-max / recombinator

🧪 Recombinator

Purpose

Produce offspring genomes from selected parent pairs via simulated meiotic
recombination. Models Mendelian segregation, de novo mutation, sex determination,
and clinical evaluation against a disease registry.

How It Works

1. Mendelian segregation: one allele inherited from each parent per locus
   (random selection simulating independent assortment).

2. De novo mutation: configurable rate per locus (default 0.1%), with hotspot
   multipliers for cognitive, immune, and metabolic loci. Mutations are classified
   as disease-risk, protective, or neutral.

3. Sex determination: 50/50 coin flip (XY or XX).

4. Trait inference: reverse-map offspring genotype back to trait scores using
   the trait registry, accounting for dominance models.

5. Clinical evaluation: check offspring genotype against disease registry for
   penetrance, onset probability, and fitness cost.

6. Health score: computed from cumulative fitness costs of clinical conditions.

Input

• Two parent .genome.json files (one Male, one Female)
• GENOMEBOOK/DATA/trait_registry.json
• GENOMEBOOK/DATA/disease_registry.json

Output

• Offspring .genome.json with:

  • Inherited loci and alleles
  • Mutation log
  • Inferred trait scores
  • Clinical history
  • Health score (0.0 to 1.0)

CLI Usage

# Demo: breed Einstein x Anning, produce 3 offspring
python skills/recombinator/recombinator.py --demo

# Breed specific parents
python skills/recombinator/recombinator.py \
  --father einstein-g0 --mother anning-g0 --offspring 3

# Custom generation number
python skills/recombinator/recombinator.py \
  --father einstein-g0 --mother curie-g0 --offspring 2 --generation 1

Output Format

ID:     g1-001-a3f2c1
Sex:    Female (XX)
Health: 0.9500
Mutations: 1
  - COMT_Val158Met: G->A (neutral, from mother)
Conditions: 0
Top traits:
  - curiosity: 0.92
  - analytical_thinking: 0.88
  - persistence: 0.85