CatMOF

We are generating DFT-ready SBU clusters for stable MOFs with catalytic potential.

Installation

From a fresh clone, use the pinned conda environment and install the package in editable mode:

git clone <repo-url>
cd CatMOF
conda env create -f devtools/conda-envs/catmof_env.yaml
conda activate catmof
pip install -e .

Then copy config.example.yaml to config.yaml, set paths (e.g. zeo_network if using geometric featurization), and run with catmof run -c config.yaml.

Text mining

CatMOF does not ship a manuscript CSV. You provide it and point config at it via paths.manuscript_data_csv. The CSV must have columns name (MOF refcode, matching stable MOF names), doi, and title.

• Title-only (text_mining_title_only: true): Provide the manuscript CSV with names, DOIs, and titles. Use the title fetcher to fill titles from a DOI list (same CSV format as the downloader):

  pip install catmof[text_mining_titles]
  python -m catmof.text_mining_tools.title_fetcher --doi-csv dois.csv --output-csv manuscript_data_w_titles.csv

  If your DOI CSV includes a name column (e.g. from a CSD export), the output is ready for manuscript_data_csv.

• Full-paper: Provide the same manuscript CSV and a directory of article HTML/XML files (paths.text_mining_html_dir). CatMOF runs the internal pickler on your HTMLs and writes pickles under the text_mining directory (existing pickles are skipped by default). To download HTML/XML by DOI first:

  python -m catmof.text_mining_tools.downloader --doi-csv dois.csv --output-dir /path/to/html/corpus

  Downloader CSV: doi column required; optional mode column. For Elsevier DOIs you need an API key (obtain separately; use --elsevier-key or CATMOF_ELSEVIER_API_KEY). ACS is skipped (violation of ACS policy). Wiley is supported.

R-factor check (SBU analysis)

If you set run_rfactor_check: true, you must provide a CSV with MOF refcodes and R-factors (paths.core_rfactors_csv). CatMOF does not ship this file. You can obtain R-factor data separately using the CCDC/CSD API (a CSD license is required). The CSV must have columns: name (MOF refcode), R-factor (numeric).

Zeo++ (geometric analysis)

CatMOF does not ship Zeo++. For geometric featurization (pore diameter, surface area, pore volume, etc.) you must download and build Zeo++ yourself, then set zeo_network in your config to the path of the network binary. Use version 0.3 to match the code paths used in CatMOF.

• Download: Zeo++ — Download

Copyright

Copyright (c) 2024, Husain Adamji

Acknowledgements

Project based on the Computational Molecular Science Python Cookiecutter version 1.1.