{"success":true,"data":{"author":"Ahmed lab, University of Michigan; derived from OMol25 (Meta FAIR Chemistry)","bucketAccessType":"public","bucketName":"ahmed-lab-umich/Organic-OMol1","bucketPrefix":"","category":"Chemistry & Materials","cloudProvider":"huggingface","createdAt":"2026-11-16T02:38:31.344Z","description":"The organic subset of the OMol-1 Train All split of Meta's Open Molecules 2025 (OMol25), republished by the Ahmed lab at the University of Michigan with molecular identifiers added. OMol25 ships coordinates, wB97M-V/def2-TZVPD energies and forces without SMILES, InChI or InChIKey; this release assigns them so structures can be grouped by molecule, split by molecule to avoid train-test leakage, and joined to experimental measurements and compound databases. It contains 24,023,732 structures of 1,090,506 molecules (about 22 structures per molecule), restricted to single-component, carbon-containing, neutral, closed-shell singlets with only B, Br, C, Cl, F, H, I, N, O, P and S. Identifiers include skeleton and stereo InChIKeys; the card advises joining on inchikey_skeleton when the external source lacks consistent stereochemistry and never on formula. Files are 200 Parquet shards under structures/, about 52 GB (measured 2026-11-15), readable with pyarrow, polars or Hugging Face datasets. Released under CC-BY-4.0, as are the OMol25 records it derives from; cite Levine et al., arXiv:2505.08762, and the Organic-OMol1 repository.","displayTitle":"Organic-OMol1: 24 Million Organic OMol25 Structures with SMILES and InChIKeys","id":"9503d988-2d1b-59db-8efa-6059e2295283","license":"CC-BY-4.0","licenseEvidenceUrl":"https://huggingface.co/datasets/ahmed-lab-umich/Organic-OMol1","objectCountEstimate":202,"probeStatus":"ok","referenceUrl":"https://arxiv.org/abs/2505.08762","region":"us","schemaCard":{"accessHints":[{"language":"python-datasets","snippet":"from datasets import load_dataset\nds = load_dataset(\"ahmed-lab-umich/Organic-OMol1\")"},{"language":"python-hub","snippet":"from huggingface_hub import snapshot_download\nsnapshot_download(repo_id=\"ahmed-lab-umich/Organic-OMol1\", repo_type=\"dataset\")"}],"caveats":["HuggingFace metadata: author=ahmed-lab-umich, license:cc-by-4.0, size_categories:10M<n<100M, format:parquet, downloads=310"],"format":"mixed","generatedAt":"2026-11-16T02:38:37.455Z","layout":{"example":"structures/records_train_data0000.parquet","pattern":"{repoId}/{filename}"},"mountPath":"/mnt/archive","sampleHeaders":[{"bytes":"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","encoding":"hex","key":"structures/records_train_data0000.parquet","truncatedAt":512}],"sampleObjects":["structures/records_train_data0000.parquet","structures/records_train_data0001.parquet","structures/records_train_data0002.parquet","structures/records_train_data0003.parquet","structures/records_train_data0004.parquet"],"sidecars":[{"bytes":"---\nlicense: cc-by-4.0\ntask_categories:\n  - tabular-regression\n  - graph-ml\ntags:\n  - chemistry\n  - quantum-chemistry\n  - molecular-geometry\n  - dft\n  - 3d\nsource_datasets:\n  - extended\nsize_categories:\n  - 10M<n<100M\nconfigs:\n  - config_name: default\n    data_files: structures/*.parquet\n---\n\n# Organic-OMol1\n\nThe organic subset of the **OMol-1 Train All split** of Meta's Open Molecules\n2025 ([facebook/OMol25](https://huggingface.co/facebook/OMol25)), with molecular\nidentifiers added. OMol25 ships coordinates, energies and forces without SMILES,\nInChI or InChIKey, so its records cannot be searched or joined by compound.\nThe records here carry those identifiers.\n\n**24,023,732 structures containing only B, Br, C, Cl, F, H, I, N, O, P and S.\nThese are 1,090,506 molecules, 22.03 structures per molecule on average.**\nEvery structure is a single-component, carbon-containing, neutral, closed-shell\nsinglet.\n\n## Questions and feedback\n\nHave a question, found an issue, or used Organic-OMol1 in your research?\nWe welcome your feedback and would love to hear about your work.\n\n[Join the discussion](https://huggingface.co/datasets/ahmed-lab-umich/Organic-OMol1/discussions).\n\n## Uses\n\nWhat the identifiers let you do that OMol25 alone does not:\n\n- **Group every structure of a molecule**, and split by molecule so that\n  structures of the same molecule do not land on both sides of a train and test\n  split.\n- **Join DFT records to experimental measurements and public compound\n  databases.** Join on `inchikey_skeleton` when the experimental source has no\n  consistent stereochemistry, which is the common case. Join on\n  `inchikey_stereo` when stereochemistry matters, at lower coverage.\n- **Do not join on `formula`.** Constitutional isomers share a formula. Formula\n  matching was tested on this corpus and fails.\n- **Train interatomic potentials and 3D property models** on coordinates,\n  energies and forces evaluated at one level of theory.\n- **Featurize the same records as SMILES or molecular graphs** for models that\n  do not take coordinates, including sequence and graph architectures.\n- **Select configurations by force** before training, using `forces_flat`.\n\nA Python package for compound matching and molecular feature generation against\nthis release is planned as a separate public repository.\n\n## Chemical identifiers added\n\n| column | what it holds |\n|---|---|\n| `canonical_smiles` | canonical isomeric SMILES |\n| `smiles_flat` | SMILES without stereochemistry |\n| `smiles_stereo` | SMILES with stereochemistry |\n| `inchi` | Standard InChI |\n| `inchikey_stereo` | full Standard InChIKey. Encodes isotope, charge and protonation state. Does not encode spin |\n| `inchikey_skeleton` | first InChIKey block, the connectivity skeleton. Collapses stereoisomers, and can collapse mobile-hydrogen tautomers |\n| `inchikey_stereo_geometric` | full Standard InChIKey with stereochemistry taken from the coordinates instead of the graph |\n| `formula` | molecular formula |\n\n## Meta Data\n\n111 in all, counted from the released Parquet schema. 82 were produced by this\nwork and do not exist in OMol25. The other 29 carry OMol25's values unchanged.\nNo new quantum chemistry was run. Every energy, force, orbital energy and\npartial charge is OMol25's calculation. A `*_status` column over an OMol25\nproperty counts as produced here, because it records whether their value is\npresent rather than holding a value of its own.\n\n| group | columns | added here | from OMol25 |\n|---|---|---|---|\n| Availability flags | 23 | 23 | 0 |\n| Identity evidence | 15 | 15 | 0 |\n| Composition and scope | 13 | 13 | 0 |\n| DFT scalars | 13 | 0 | 13 |\n| Per-atom arrays | 12 | 0 | 12 |\n| Parsed inputs with validity | 9 | 9 | 0 |\n| Record provenance | 9 | 5 | 4 |\n| Chemical identifiers | 8 | 8 | 0 |\n| Force and sampling strata | 6 | 6 | 0 |\n| Units and geometry bookkeeping | 3 | 3 | 0 |\n\nEach physical property has a `*_status` column saying whether that property is\npresent and usable for that structure. Filter on it before regressing on a value.\nA null is not a zero.\n\n## How to load it\n\nOne table, `struc","encoding":"utf8","key":"README.md","name":"README.md","size":8191,"truncated":true}],"sourceObjectKey":"structures/records_train_data0000.parquet","starterCell":"from datasets import load_dataset\nds = load_dataset(\"ahmed-lab-umich/Organic-OMol1\")\nprint(ds)","starterSamplePath":"structures/records_train_data0000.parquet","totalObjects":202},"summary":"The organic subset of the OMol-1 Train All split of Meta's Open Molecules 2025 (OMol25), republished by the Ahmed lab at the University of Michigan with molecular identifiers…","title":"Organic-OMol1: 24 Million Organic OMol25 Structures with SMILES and InChIKeys","updatedAt":"2026-11-16T17:08:56.185Z","visibility":"public","websiteUrl":"https://huggingface.co/datasets/ahmed-lab-umich/Organic-OMol1"}}