# Soul2dna

> Compiles SOUL.md character profiles into synthetic diploid genomes by mapping trait scores to alleles at defined loci, producing .genome.json files.

- Skill: `comeonoliver/soul2dna` (Agent Skill)
- Install (CLI): `npx skillmds add comeonoliver/soul2dna`
- Raw SKILL.md: https://api.skillmd.com/api/skills/comeonoliver/soul2dna/raw
- Safety review: PASS (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML, Data & Analytics, Model Training & Fine-tuning
- Tags: Genetics, Genome, Json, Soul Md, Trait Mapping
- License: MIT
- Author: ComeOnOliver (https://skillmd.com/u/comeonoliver)
- Updated: 2026-08-22
- Page: https://skillmd.com/skills/comeonoliver/soul2dna

---


# 🧬 Soul2DNA Compiler

## Purpose

Compile SOUL.md character profiles into synthetic diploid genomes. Each soul file
describes a historical or fictional figure with trait scores (0.0 to 1.0). The
compiler maps these scores to alleles at defined loci using additive, dominant, or
recessive inheritance models, producing a `.genome.json` file per character.

## How It Works

1. **Parse SOUL.md** files from `GENOMEBOOK/DATA/SOULS/` extracting identity
   metadata (name, sex, ancestry, domain, era) and trait scores.
2. **Load trait registry** (`GENOMEBOOK/DATA/trait_registry.json`) which defines
   loci, alleles, chromosomal positions, dominance models, and effect sizes for
   each trait.
3. **Assign genotypes** at each locus based on trait score thresholds:
   - Additive: <0.33 ref/ref, 0.33-0.66 ref/alt, >0.66 alt/alt
   - Dominant: <0.40 ref/ref, 0.40-0.75 ref/alt, >0.75 alt/alt
   - Recessive: <0.50 ref/ref, 0.50-0.80 ref/alt, >0.80 alt/alt
4. **Write genome** as JSON with full locus detail, trait scores, and metadata.

## Input

- `GENOMEBOOK/DATA/SOULS/*.soul.md` (20 historical figures)
- `GENOMEBOOK/DATA/trait_registry.json`

## Output

- `GENOMEBOOK/DATA/GENOMES/<name>-g0.genome.json` per character

## CLI Usage

```bash
# Compile all souls to genomes
python skills/soul2dna/soul2dna.py

# Demo mode (shows summary without writing files)
python skills/soul2dna/soul2dna.py --demo
```

## Output Format

Each `.genome.json` contains:

```json
{
  "id": "einstein-g0",
  "name": "Albert Einstein",
  "sex": "Male",
  "sex_chromosomes": "XY",
  "ancestry": "...",
  "generation": 0,
  "parents": [null, null],
  "loci": { "<locus_id>": { "chromosome": "...", "alleles": ["A","G"], ... } },
  "trait_scores": { "curiosity": 0.95, ... }
}
```