MofasaDB: 200,000 Generated Metal-Organic Frameworks from the Mofasa Diffusion Model
- Provider
- Orbital Materials
- License
- CC-BY-4.0 (source)
- Size
- 62 files
- Format
- NumPy (in sampled files)
- Stored at
- Hugging Face (https://huggingface.co/datasets/Orbital-Materials/MofasaDB)
- Access
- Free and public. No account with the data provider is needed.
- Category
- Chemistry & Materials
- Source
- https://mofux.ai/MOFASA.pdf
About this dataset
MofasaDB is the structure database released by Orbital Materials with Mofasa, a generative diffusion model for metal-organic frameworks trained on QMOF structures of up to 170 atoms. It holds more than 200,000 unconditionally generated de novo MOFs in samples.db and their geometry-relaxed counterparts in relaxed.db (ASE SQLite databases, row-aligned and 1-indexed), plus per-structure ORB graph latent embeddings as NumPy arrays under sample_latents/ and relaxed_latents/, about 18.6 GB in total (measured 2026-11-15). Each row carries ORB potential energy per atom and maximum force, Zeo++ geometric properties (largest cavity and pore-limiting diameters, accessible volume and surface area, channels and pockets), space group, MOFid and MOFkey with topology, node and linker SMILES, MOFChecker validity flags, fragment and linker descriptors, and validation metrics. Read structures with ase.db.connect and row.toatoms(); a search interface is at mofux.ai. Used for generative model evaluation, MOF discovery, porosity screening and representation learning. Released by Orbital Materials under CC-BY-4.0; cite the Mofasa paper (Orbital Materials, 2025).
How to load this dataset in Python
from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="Orbital-Materials/MofasaDB", filename="relaxed_latents/linkers_morgan_ecfp4.npy", repo_type="dataset")
print(path)Data structure
File types in sampled files: NumPy
Files
Example file: relaxed_latents/linkers_morgan_ecfp4.npy
Analyze this dataset without downloading it
On Scigantic, MofasaDB: 200,000 Generated Metal-Organic Frameworks from the Mofasa Diffusion Model opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.
This page is also available as Markdown and as JSON.
Related Chemistry & Materials datasets
- Halide Perovskite Ion Migration MLFF positive iodide vacancy
- Iron Nanoparticle DFT Database: Clusters, Bulk Phases, Surfaces and Liquid Fe (Jana and Caro 2023), ColabFit Conversion
- Open Molecules 2025 (OMol25) 4M Training Subset, ColabFit Conversion
- MassBank 2026.03, open-licence records: 96,949 reference mass spectra of 10,304 compounds as MSP and Parquet
- CoRE MOF 2024: Computation-Ready Experimental Metal-Organic Frameworks (Public SI Subset)
- Open Catalyst 2020 (OC20) S2EF 200K Training Set, ColabFit Conversion
- Stable and Metastable CuInS2 (CIS) Chalcopyrite Polytypes for Solar Cells: DFT Formation Energies (Larsen et al. 2022, Materials Cloud), ColabFit Conversion
- JARVIS-ML CFID Descriptors for 84K Materials Project Crystals: Formation Energies (NIST JARVIS, figshare), ColabFit Conversion