CoRE MOF Tools

CoRE MOF Tools is the Python interface accompanying the CoRE MOF database. It provides database lookup and structure download, CIF curation and validation, geometric and chemical descriptors, and pretrained property models.

The package combines pure-Python functions with optional external software. Run the installation check before starting:

coremof doctor
# or: python -m CoREMOF doctor

Installation

Python 3.9–3.11 is supported. A clean conda environment is recommended because several scientific dependencies contain compiled extensions.

conda create -n coremof python=3.11
conda activate coremof
pip install CoREMOF-tools
coremof doctor

To reproduce the repository environment, use:

conda env create -f env.yaml
conda activate coremof_tools

Some features require additional software:

Feature	Additional requirement
Zeo++ pore geometry	conda install -c conda-forge zeopp-lsmo
CSD structure download and MOSAEC	Licensed CSD software and CSD Python API
MOFChecker	pip install git+https://github.com/Au-4/mofchecker_2.0.git@main
MOFid v1/v2	MOFid installation and Open Babel
Crystal topology	Julia/CrystalNets through juliacall; the first call may install Julia packages
Heat-capacity prediction	Full repository checkout containing the ~1.3 GB ensemble model directory

Quick start

Query a CoRE MOF record

from CoREMOF.structure import information

record = information("CR-ASR", "2020[Cu][sql]2[ASR]1")
print(record)

Valid datasets are CR-ASR, CR-FSR, CR-Ion, and NCR. Invalid dataset or entry names now raise an error with valid choices or close matches.

Calculate pore geometry with Zeo++

from CoREMOF.calculation import Zeopp

diameters = Zeopp.PoreDiameter("my_mof.cif")
print(diameters["LCD"], diameters["PLD"], diameters["unit"])

surface_area = Zeopp.SurfaceArea(
    "my_mof.cif",
    chan_radius=1.86,
    probe_radius=1.86,
    num_samples=10_000,
)
print(surface_area["ASA"])

Zeo++ calls use unique temporary files and are safe for paths containing spaces and for concurrent workflows. Set COREMOF_NETWORK_EXECUTABLE if the executable is not named network.

Precheck and standardize a CIF

from CoREMOF.curate import preprocess

job = preprocess("my_mof.cif", output_folder="result_curation")
print(job.result_check)

The curation classes write CIF/JSON/CSV outputs to the requested output directory. They do not return the processed CIF in memory.

Predict properties

from CoREMOF.prediction import pacman, stability

charge_result = pacman("my_mof.cif", output_folder="result_pacman")
stability_result = stability("my_mof.cif")

stability() additionally calls Zeo++ and RAC descriptors. Check all requirements with coremof doctor before running it.

Feature map

Task	Module	Main entry point
Database lookup	CoREMOF.structure	information()
SI/CSD structure download	CoREMOF.structure	download_from_SI, download_from_CSD()
Zeo++ geometry	CoREMOF.calculation.Zeopp	PoreDiameter(), SurfaceArea(), PoreVolume()
Basic descriptors	CoREMOF.calculation.mof_features	SpaceGroup(), Mass(), Volume(), n_atom()
Topology, OMS, RACs	CoREMOF.calculation.mof_features	topology(), get_oms_file(), RACs()
CIF curation	CoREMOF.curate	preprocess, clean, clean_pacman
Structural validation	CoREMOF.curate	mof_check, run_MOSAEC(), run_mofclassifier()
Property models	CoREMOF.prediction	pacman(), cp(), stability()
MOF identifiers	CoREMOF.get_mofid	run_v1(), run_v2()

Full guides and API documentation are available on Read the Docs. Executable notebooks and CIF examples are in examples/.

Citation

If you use the database or toolkit, cite:

Zhao G., Brabson L., Chheda S., Huang J., Kim H., Liu K., et al. “CoRE MOF DB: a curated experimental metal–organic framework database with machine-learned properties for integrated material-process screening.” Matter 8 (2025), 102140. https://doi.org/10.1016/j.matt.2025.102140

Also cite the underlying method used in your workflow (for example Zeo++, PACMAN-charge, MOFClassifier, MOFChecker, MOSAEC, CrystalNets, or the relevant stability model). A method-by-method list is provided in the documentation.

Support and development

• Report reproducible bugs through GitHub Issues.
• Include coremof doctor output, Python version, operating system, a minimal code example, and—when shareable—the failing CIF.
• Developed by Guobin Zhao at MTAP, Pusan National University.