# Rdkit

> RDKit cheminformatics via VenusFactory bioinfo scripts and agent_generated_code. Use for SMILES/SDF, descriptors, fingerprints, substructure filters, similarity. Do NOT use for ChEMBL bioactivity download (chembl_database) or openFDA (fda). No dedicated rdkit_* LangChain tool — run scripts or import modules.

- Skill: `ai4protein/rdkit` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx skillmds@latest add ai4protein/rdkit`
- Raw SKILL.md: https://api.skillmd.com/api/skills/ai4protein/rdkit/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- License: BSD-3-Clause
- Author: ai4protein (https://skillmd.com/u/ai4protein)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/ai4protein/rdkit

---


# RDKit Cheminformatics

## Overview

Fine-grained molecular control in Python. Platform helpers live under `src/tools/bioinfo/rdkit/` (CLI + importable functions). Prefer them over re-implementing RDKit from scratch.

## VenusFactory execution

- No `rdkit_*` `@tool` in the hub — use **`agent_generated_code`** (or `python_repl`) and import the modules below.
- Extended tutorials: `read_skill("rdkit", relative_path="references/legacy_guide.md")`.
- Bioactivity from ChEMBL first → `chembl_database`, then RDKit for chemistry ops.

## Project scripts (src/tools/bioinfo/rdkit/)

| Script | Purpose |
|--------|---------|
| `molecular_properties.py` | MW, LogP, TPSA, Lipinski, QED → CSV |
| `substructure_filter.py` | SMARTS/SMILES include/exclude filters |
| `similarity_search.py` | Morgan/RDKit/MACCS fingerprints + Tanimoto |

```python
from src.tools.bioinfo.rdkit import calculate_properties, filter_molecules, similarity_search
# loaders: substructure_filter.load_molecules / similarity_search.load_molecules
```

```bash
python -m src.tools.bioinfo.rdkit.molecular_properties "CCO"
python -m src.tools.bioinfo.rdkit.similarity_search "CCO" database.smi --threshold 0.7 -o hits.csv
```

## When NOT to use

- Download ChEMBL molecules/activities → `chembl_database`
- Simple FDA label search → `fda` / `query_fda`

## Common mistakes

- Importing `load_molecules` from package root (use module-specific loaders)
- Assuming a LangChain tool named `rdkit_*` exists
- Sanitization failures on dirty SMILES — handle `None` mols

## References

- `references/api_reference.md`, `descriptors_reference.md`, `smarts_patterns.md`
- `references/legacy_guide.md` — full capability cookbook