PubChemQC PM6: Semi-Empirical Geometries and Properties for 221 Million Calculations
- Provider
- Maho Nakata et al. (RIKEN PubChemQC project); packaged by MolSSI
- License
- Public Domain (source)
- Size
- 2,426 files
- Format
- JSON (in sampled files)
- Stored at
- Hugging Face (https://huggingface.co/datasets/molssiai-hub/pubchemqc-pm6)
- Access
- Free and public. No account with the data provider is needed.
- Category
- Chemistry & Materials
- Source
- https://nakatamaho.riken.jp/pubchemqc.riken.jp/
About this dataset
The PubChemQC PM6 data set from Maho Nakata's PubChemQC project at RIKEN, packaged by MolSSI: optimised molecular geometries and electronic properties computed with the PM6 semi-empirical method for 94.0% of the 91.6 million molecules in PubChem Compound as retrieved on 29 August 2016. Besides neutral states, cationic, anionic and spin-flipped states were computed for 56.2%, 49.7% and 41.3% of the molecules respectively, giving 221 million PM6 calculations in total. Measured 2026-11-14, data/ holds seven configurations: the full pm6opt set and filtered subsets by element set and size (for example pm6opt_chon300nosalt and pm6opt_chnopsfclnakmgca500); the repository uses about 8.4 TB. Load with Hugging Face datasets (streaming advisable) or read the shards directly. Useful for pretraining on charged and open-shell states, ionization and electron affinity models. The RIKEN publisher states that PubChemQC results are in the public domain; the MolSSI card distributes them under CC-BY-4.0. Cite Nakata et al., JCIM 2020.
How to load this dataset in Python
from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="molssiai-hub/pubchemqc-pm6", filename="pubchemqc-pm6.py", repo_type="dataset")
print(path)Data structure
File types in sampled files: JSON
Files
Example file: pubchemqc-pm6.py
Analyze this dataset without downloading it
On Scigantic, PubChemQC PM6: Semi-Empirical Geometries and Properties for 221 Million Calculations 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
- Generalized Stacking Fault Energies (GSFE) of Al, Cu and Mg under Normal Stress: DFT (Yin, Andric and Curtin 2019, Materials Cloud), ColabFit Conversion
- Zeo-1: DFT Energies and Stresses for Zeolite Frameworks from the IZA Database, ColabFit Conversion
- ChEBI: the ontology of chemical entities of biological interest, with structures and cross-references
- GST-GAP-22: Ge-Sb-Te Phase-Change Memory Material DFT Training Set, ColabFit Conversion
- Amorphous Lithium-Silicon (LixSi) Anode Structures for aenet Potentials, ColabFit Conversion
- mbGDML: Water, Acetonitrile and Methanol Cluster MD with Many-Body Machine Learning Potentials, ColabFit Conversion
- Co(II) Dimer Single-Molecule Magnet Ligand Substitutions: Exchange Energy DFT (Ren et al. 2022, Materials Cloud), ColabFit Conversion
- MPtrj Relaxation Trajectories and Materials Project 2023 Structures (Matbench Discovery Files)