{"success":true,"data":{"author":"Orbital Materials","bucketAccessType":"public","bucketName":"Orbital-Materials/MofasaDB","bucketPrefix":"","category":"Chemistry & Materials","cloudProvider":"huggingface","createdAt":"2026-11-16T02:38:31.173Z","description":"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).","displayTitle":"MofasaDB: 200,000 Generated Metal-Organic Frameworks from the Mofasa Diffusion Model","id":"e792a200-d36b-5a7a-a1c1-8ca4d6c75294","license":"CC-BY-4.0","licenseEvidenceUrl":"https://huggingface.co/datasets/Orbital-Materials/MofasaDB","objectCountEstimate":62,"probeStatus":"ok","referenceUrl":"https://mofux.ai/MOFASA.pdf","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"Orbital-Materials/MofasaDB\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"Orbital-Materials/MofasaDB\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=Orbital-Materials, license:cc-by-4.0, size_categories:100K<n<1M, downloads=237"],"format":"mixed","generatedAt":"2026-11-16T02:38:37.167Z","layout":{"example":"relaxed_latents/linkers_morgan_ecfp4.npy","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","decoded":{"fields":{"descr":"|u1","dtype":"uint8","fortranOrder":false,"ndim":2,"shape":"201926,2048"},"format":"npy","summary":"NPY uint8 [201926, 2048]"},"encoding":"hex","key":"relaxed_latents/linkers_morgan_ecfp4.npy","truncatedAt":512}],"sampleObjects":["relaxed.db","relaxed_latents/linkers_morgan_ecfp4.npy","relaxed_latents/linkers_morgan_ecfp4_standardized.npy","relaxed_latents/linkers_morgan_ecfp6.npy","relaxed_latents/linkers_morgan_ecfp6_standardized.npy"],"sidecars":[{"bytes":"---\nlicense: cc-by-4.0\ntags:\n- chemistry\n- materials\n- diffusion\n- synthetic\npretty_name: MofasaDB\nsize_categories:\n- 100K<n<1M\n---\n# MofasaDB\n\n\nThe MofasaDB is a publicly available dataset containing 200.000+ _de novo_ generated MOF (Metal-Organic Framework) structures from Mofasa trained on QMOF (up to 170 atoms), along with their geometry-relaxed counterparts. The database is released alongside the paper [Mofasa: A Step Change in Metal-Organic Framework Generation](https://mofux.ai/MOFASA.pdf). A user-friendly web interface for search and discovery can be accessed at [https://mofux.ai/](https://mofux.ai/).\n\n---\n\n## Table of Contents\n\n1. [Database Overview](#database-overview)\n2. [Quick Start](#quick-start)\n3. [Property Reference](#property-reference)\n4. [Structural Properties](#structural-properties)\n5. [MOFID Properties](#mofid-properties)\n6. [Zeo++ Geometric Properties](#zeo-geometric-properties)\n7. [ORB Properties](#orb-properties)\n8. [MOFChecker Properties](#mofchecker-properties)\n9. [MOF Fragment Properties](#mof-fragment-properties)\n10. [Linker Properties](#linker-properties)\n11. [Validation Metrics](#validation-metrics)\n\n---\n\n## Database Overview\n\nThe database contains unconditionally generated MOF structures from Mofasa, along with their geometry-relaxed counterparts.\n\n### Files\n\n| File | Description |\n|------|-------------|\n| `samples.db` | Original generated MOF structures |\n| `relaxed.db` | Geometry-relaxed versions of the samples |\n| `sample_latents/` | ORB latent embeddings for samples |\n| `relaxed_latents/` | ORB latent embeddings for relaxed structures |\n\n### Data Alignment\n\nThe databases are **row-aligned**: row *i* in `samples.db` corresponds to row *i* in `relaxed.db`.\n\n**Indexing:**\n- ASE databases are **1-indexed**: first row is `db.get(1)`\n- NumPy arrays are **0-indexed**: first element is `array[0]`\n- Therefore: `latent[i]` corresponds to `db.get(i + 1)`\n\n---\n\n## Quick Start\n\n### Load a Structure\n\n```python\nfrom ase.db import connect\n\ndb = connect(\"samples.db\")\nrow_id = 1\nrow = db.get(row_id)                    # Get first structure (1-indexed)\natoms = row.toatoms()              # Convert to ASE Atoms object\nprint(atoms.get_chemical_formula())\n```\n\n### Access Properties\n\n```python\n# Get energy per atom\nenergy = row.data['properties']['orb_properties']['orb_energy_per_atom']\n\n# Get pore diameter\nlcd = row.data['properties']['pyzeo_geometric_properties']['lcd']\n\n# Get topology (top-level property)\ntopology = row.data['topology']\n```\n\n### Load Orb Latent Embeddings\n\n```python\nimport numpy as np\n\nlatents = np.load(\"sample_latents/orb_latent_4_graph.npy\")\nlatent = latents[row_id - 1]  # Convert 1-indexed row to 0-indexed array\n```\n\n### Compare Sample and Relaxed\n\n```python\nsample_db = connect(\"samples.db\")\nrelaxed_db = connect(\"relaxed.db\")\n\n# Row i in both databases correspond to the same structure\nrow_id = 100\nsample_atoms = sample_db.get(row_id).toatoms()\nrelaxed_atoms = relaxed_db.get(row_id).toatoms()\n\nprint(f\"Sample formula:  {sample_atoms.get_chemical_formula()}\")\nprint(f\"Relaxed formula: {relaxed_atoms.get_chemical_formula()}\")\n```\n\n### Handle Missing Data\n\nNot all properties are available for every structure. Common causes include:\n\n- **MOFID failure**: If MOFID cannot identify the MOF building blocks (nodes, linkers, topology), these properties are set to `\"UNKNOWN\"`, `\"ERROR\"`, or empty lists for missing SMILES strings. \n- **Zeo++ non-porous**: If Zeo++ determines a structure has insufficient porosity for probe access, geometric properties (`lcd`, `pld`, accessible volume/surface area) may be missing, zero, or `None`.\n- **Component absence**: Latent embeddings for `bound_solvent` and `free_solvent` are zero vectors when structures contain no solvent molecules.\n\n---\n\n## Property Reference\n\nProperties are stored in `row.data` with nested paths. Some examples:\n\n```python\nPROPERTY_PATHS = {\n    # ORB model properties\n    'orb_energy_per_atom': 'properties.orb_properties.orb_energy_per_atom',\n    'orb_max_force': 'properties.orb_properties.orb_max_force',\n    \n    # Zeo++ geometric properties\n  ","encoding":"utf8","key":"README.md","name":"README.md","size":20243,"truncated":true}],"sourceObjectKey":"relaxed_latents/linkers_morgan_ecfp4.npy","starterCell":"from huggingface_hub import hf_hub_download\npath = hf_hub_download(repo_id=\"Orbital-Materials/MofasaDB\", filename=\"relaxed_latents/linkers_morgan_ecfp4.npy\", repo_type=\"dataset\")\nprint(path)","starterSamplePath":"relaxed_latents/linkers_morgan_ecfp4.npy","totalObjects":62},"summary":"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…","title":"MofasaDB: 200,000 Generated Metal-Organic Frameworks from the Mofasa Diffusion Model","updatedAt":"2026-11-16T16:51:49.398Z","visibility":"public","websiteUrl":"https://mofux.ai/"}}