# Analyze Molecules with RDKit

Advanced molecular workflows require careful control over parsing, descriptors, fingerprints, and structure queries. This skill provides focused RDKit guidance, references, and reusable scripts.

## Install

```bash
npx skillstore add davila7/rdkit
```

## Metadata

- Status: approved
- Slug: davila7-rdkit
- Skillstore revision: r2
- Version status: missing
- Tree hash: 44b5bf8574ec3901d9af69cfad5a30a9698b5b3ec404f7526b3de9c62df03f83
- Author: davila7
- GitHub username: davila7
- License: MIT
- Repository: https://github.com/davila7/claude-code-templates/tree/main/cli-tool/components/skills/scientific/rdkit
- Ref: c43861a65bb95efcae259cd161c9d6f4dc7eec6f
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: confirmation\_required
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: filesystem, scripts, external\_commands
- Quality score: 38
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/davila7-rdkit
- Manifest: https://skillstore.pages.dev/api/skills/davila7-rdkit/manifest

## Capabilities

- Parse SMILES, SDF, MOL, InChI, and molecule blocks with validation guidance.
- Calculate molecular weight, LogP, TPSA, hydrogen-bond counts, QED, and other descriptors.
- Generate molecular fingerprints and compare structures with Tanimoto, Dice, or cosine similarity.
- Filter molecule collections with custom SMARTS patterns and predefined structural libraries.
- Generate 2D depictions, 3D conformers, alignments, and highlighted substructure images.
- Guide reaction SMARTS, molecular modification, standardization, and scaffold analysis.

## Use Cases

- Screen Compound Libraries: Calculate properties, apply structural filters, and rank compounds by fingerprint similarity.
- Prepare Molecular Features: Create validated descriptors and fingerprints for modeling, clustering, or exploratory analysis.
- Build Reproducible Structure Workflows: Design parsing, sanitization, conformer generation, reaction, and visualization workflows with explicit parameters.

## Prompt Templates

### Calculate Basic Properties

```
Analyze the SMILES CCO. Report molecular weight, LogP, TPSA, hydrogen-bond donors, hydrogen-bond acceptors, and Lipinski status.
```

### Prepare a Descriptor Export

```
Review molecules.smi and prepare a command to calculate descriptors into properties.csv. Explain how invalid molecules are handled.
```

### Run a Similarity Search

```
Search database.sdf for compounds similar to c1ccccc1 using Morgan fingerprints and a 0.70 Tanimoto threshold. Summarize the top hits.
```

### Design an Advanced Pipeline

```
Design an RDKit workflow that filters PAINS, computes custom descriptors, clusters Morgan fingerprints, generates conformers, and records validation failures.
```

## Limitations

- Python and RDKit must be installed separately before running the examples or scripts.
- The included scripts support common molecular formats but do not cover every RDKit data source.
- Descriptor values and structure matches require scientific review for each research context.
- 3D conformers and force-field results depend on molecular preparation, parameters, and random seeds.

## Best Practices

- Check every parsed molecule for None before calculating properties or fingerprints.
- Validate SMARTS patterns and document fingerprint parameters, thresholds, and random seeds.
- Use exchange formats such as SDF or SMILES for untrusted data instead of pickle files.

## Anti Patterns

- Do not process invalid molecules without checking parser results.
- Do not compare similarity scores generated with different fingerprint settings.
- Do not load pickle files from untrusted or unknown sources.

## Security Audit

- Audited at: 2026-07-23T14:02:11.315\+00:00
- Summary: All 115 static findings are false positives after contextual review. They represent Markdown syntax, chemistry notation, expected output writes, dictionary keys, or normal error messages. A separate high-severity issue remains because SKILL.md demonstrates unsafe pickle deserialization without a trust warning.

## Stats

- Views: 244
- Downloads: 9
- Favorites: 0
- Popularity score: 0