PubChemQC B3LYP/6-31G*//PM6: Electronic Properties of 86 Million PubChem Molecules
- Provider
- Maho Nakata et al. (RIKEN PubChemQC project); packaged by MolSSI
- License
- Public Domain (source)
- Size
- 1,587 files
- Format
- JSON (in sampled files)
- Stored at
- Hugging Face (https://huggingface.co/datasets/molssiai-hub/pubchemqc-b3lyp)
- Access
- Free and public. No account with the data provider is needed.
- Category
- Chemistry & Materials
- Source
- https://doi.org/10.1021/acs.jcim.3c00899
About this dataset
The PubChemQC B3LYP/6-31G*//PM6 data set from Maho Nakata's PubChemQC project at RIKEN, packaged by the Molecular Sciences Software Institute (MolSSI). It reports electronic properties for 85,938,443 molecules, about 94% of the PubChem Compound catalog as of 29 August 2016 with molecular weight up to 1000: properties computed with the B3LYP/6-31G* and PM6 methods, including orbitals, orbital energies, total energies, dipole moments and related properties. Measured 2026-11-14, data/ holds six configurations: the full set (b3lyp_pm6) and filtered subsets by element set and size (chon300nosalt, chon500nosalt, chnopsfcl300nosalt, chnopsfcl500nosalt, chnopsfclnakmgca500), stored as JSON shards under train/; the repository uses about 7.7 TB. Load a configuration with Hugging Face datasets (streaming is advisable) or read the JSON directly. It is used for large-scale pretraining of molecular property models and HOMO-LUMO gap prediction. The RIKEN publisher states that the results are in the public domain; the MolSSI card distributes them under CC-BY-4.0. Cite Nakata et al., JCIM 2023.
How to load this dataset in Python
from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="molssiai-hub/pubchemqc-b3lyp", filename="pubchemqc-b3lyp.py", repo_type="dataset")
print(path)Data structure
File types in sampled files: JSON
Files
Example file: pubchemqc-b3lyp.py
Analyze this dataset without downloading it
On Scigantic, PubChemQC B3LYP/6-31G*//PM6: Electronic Properties of 86 Million PubChem Molecules 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
- ANI-1x Active Learning Molecular Conformers (wB97x/6-31G*), ColabFit Conversion
- HEA25: High-Entropy Transition-Metal Alloy DFT Training Set, ColabFit Conversion
- Calcium Ferrites CaFe(2+n)O(4+n) and Ca4Fe9O17 as Calcium-Ion Battery Cathodes: DFT (Arroyo-de Dompablo 2020, Materials Cloud), ColabFit Conversion
- COLL Molecular Collisions Dataset (Training Split), ColabFit Conversion
- AIMNet2 (2025) Noncovalent Interaction Training Data (B97-3c), ColabFit Conversion
- Alexandria PBE 2D Crystal Geometry Optimization Paths, ColabFit Conversion
- Materials Project Elastic Tensors (2024-11-14 Release)
- ANI-2x Conformers Recomputed at wB97M-D3(BJ)/def2-TZVPP, ColabFit Conversion