# Daylight Theory

> Daylight Theory — Cheminformatics Fundamentals

- Skill: `kdevos12/daylight-theory` (Agent Skill, multi-file: 7 files)
- Install (CLI): `npx skillmds@latest add kdevos12/daylight-theory`
- Raw SKILL.md: https://api.skillmd.com/api/skills/kdevos12/daylight-theory/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: Kdevos12 (https://skillmd.com/u/kdevos12)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/kdevos12/daylight-theory

---


# Daylight Theory — Cheminformatics Fundamentals

The Daylight Theory Manual is the canonical reference for the molecular languages underlying modern cheminformatics: SMILES, SMARTS, SMIRKS, and fingerprints. These are not Daylight-proprietary — they are industry-standard formats implemented in RDKit, OpenBabel, CDK, and every major cheminformatics toolkit.

## Router — What to Read

| Topic | Reference |
|-------|-----------|
| Molecular graph model, aromaticity, chirality, SSSR, reaction representation | `references/molecules.md` |
| SMILES syntax: atoms, bonds, branches, rings, stereochemistry, reactions | `references/smiles.md` |
| SMARTS query language: primitives, operators, recursive SMARTS, reaction queries | `references/smarts.md` |
| SMIRKS reaction transforms: atom maps, grammar, stereochemistry | `references/smirks.md` |
| Fingerprints: structural keys, path-based FP, folding, Tanimoto, Tversky, all similarity measures | `references/fingerprints.md` |
| Chemical database concepts: hash tables, identifiers, in-memory search, pools, hitlists | `references/cheminformatics-databases.md` |

## Core Languages at a Glance

| Language | Purpose | Example |
|----------|---------|---------|
| SMILES | Encode a specific molecule | `CC(=O)Oc1ccccc1C(=O)O` |
| SMARTS | Describe a molecular pattern | `[OH]c1ccccc1` (phenol) |
| SMIRKS | Encode a reaction transform | `[C:1][Br:2]>>[C:1][I:2]` |

## Key Principles

- **Molecular graph**: atoms = nodes, bonds = edges; properties are explicit or derived
- **Aromaticity**: Hückel 4N+2 rule, all ring atoms sp² — a derived, not stored, property
- **SMILES organic subset**: B, C, N, O, P, S, F, Cl, Br, I — no brackets needed at normal valence
- **SMARTS unspecified = unrestricted**: `O` in SMARTS matches any aliphatic oxygen; in SMILES it means water
- **Tanimoto**: c/(a+b+c) — the standard fingerprint similarity; double-zero independent

## Related ALKYL Skills

- `rdkit` — implementation of these concepts in Python
- `scientific-skills:matchms` — spectrum similarity (same mathematical ideas)