Organic-OMol1: 24 Million Organic OMol25 Structures with SMILES and InChIKeys

Provider
Ahmed lab, University of Michigan; derived from OMol25 (Meta FAIR Chemistry)
License
CC-BY-4.0 (source)
Size
202 files
Format
Parquet (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/ahmed-lab-umich/Organic-OMol1)
Access
Free and public. No account with the data provider is needed.
Category
Chemistry & Materials
Source
https://arxiv.org/abs/2505.08762

About Organic-OMol1: 24 Million Organic OMol25 Structures with SMILES and InChIKeys

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.

How to load this dataset in Python

from datasets import load_dataset
ds = load_dataset("ahmed-lab-umich/Organic-OMol1")
print(ds)

Data structure

File types in sampled files: Parquet

Files

Example file: structures/records_train_data0000.parquet

Analyze this dataset without downloading it

On Scigantic, Organic-OMol1: 24 Million Organic OMol25 Structures with SMILES and InChIKeys 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