chem-db-qmof

Files (8)

• SKILL.md
• README.md · examples/hkust1_dotsov01/
• input_configs.yaml · examples/hkust1_dotsov01/
• qmof-8b5bb88.cif · examples/hkust1_dotsov01/
• qmof_properties.json · examples/hkust1_dotsov01/
• validation.json · examples/hkust1_dotsov01/
• test_qmof.sh · examples/
• query_qmof.py · scripts/

SKILL.md (first lines)

# chem-db-qmof

## Goal

To retrieve computational data and relaxed crystal structures (.cif) for roughly 20,000 Metal-Organic Frameworks (MOFs) that have been thoroughly characterized using DFT. QMOF is hosted by the Materials Project under the "MPContribs" project (`qmof`). This skill allows you to retrieve relaxed CIF structures by reference name, CSD refcode, or specific structural/electronic properties like bandgap.

## Prerequisites

- **Dependency**: `mpcontribs-client` must be installed.
- **Environment**: Execution happens in the `cpu` environment.
- **Credentials**: Requires the standard Materials Project API key exported as `MP_API_KEY`.

## Instructions

1. **Query Database**: Use the provided script `query_qmof.py` to search by formula or identifier.

```bash
${CLAUDE_SKILL_DIR}/../../venv/run cpu python ${CLAUDE_SKILL_DIR}/scripts/query_qmof.py \
    --formula "Zn" \
    --max-results 5 \
    --output-dir ./research/qmof_results
```

### Available Arguments:
- `--formula`: Search by chemical formula or elements (e.g., `Zn,O,C`).
- `--identifier`: Search by QMOF ID (e.g., `qmof-8b5bb88`) or CSD refcode (e.g., `KAXQIL`, `DOTSOV01`). Refcodes are matched as case-sensitive substrings of the QMOF source filename. MOF common names (e.g., HKUST-1) are not indexed.
- `--max-results`: Maximum number of structures to download (default: 5).
- `--output-dir`: Directory to save the resulting `.cif` files.

Outputs: one `<qmof_id>.cif` per hit, `qmof_properties.json` (tabulated QMOF properties such as `EgPBE` band gap, density, pore sizes, source refcode/DOI, with a `units` block), and `input_configs.yaml`.

## Examples

**Literature-validated example: HKUST-1 (CSD refcode DOTSOV01).** See [examples/hkust1_dotsov01/README.md](examples/hkust1_dotsov01/README.md). It retrieves the DFT-relaxed cell and PBE band gap, and checks them against the experimental lattice parameter and the official QMOF release.

**Example 1: Automated testing script (Zinc MOF)**
```bash
bash ${CLAUDE_SKILL_DIR}/examples/test_qmof.sh
```

**Example 2: Query for 5 MOFs containing Zinc manually**
```bash
${CLAUDE_SKILL_DIR}/../../venv/run cpu python ${CLAUDE_SKILL_DIR}/scripts/query_qmof.py \
    --formula "Zn" \
    --max-results 5 \
    --output-dir ./results/qmof_zn
```

**Example 2: Retrieve a specific MOF by identifier**
```bash
${CLAUDE_SKILL_DIR}/../../venv/run cpu python ${CLAUDE_SKILL_DIR}/scripts/query_qmof.py \
    --identifier "KAXQIL" \
    --max-results 1 \
    --output-dir ./results/qmof_kaxqil
```

## Constraints

- **API Limits**: The Materials Project/MPContribs API has rate limits. Do not use extremely large values for `--max-results` unless necessary.
- **Availability**: If you cannot find a MOF in QMOF via the script, or if the API query is overly complex, you can also search the QMOF database manually through the Materials Project web interface under the "Contributions" section.

---

**Author:** Bowen Deng
**Contact:** [GitHub @learningmatter-mit](https://github.com/learningmatter-mit)

From learningmatter-mit/AtomisticSkills, MIT.

More from learningmatter-mit/AtomisticSkills