molssiai-hub/pubchem-04-18-2025

PubChem Dataset (version 04-18-2025)

Paper | Code | Project Homepage

• —Important Note (Jul. 22, 2025): According to Hugging Face, the usage of data load scripts and the corresponding trust_remote_code parameter in the load_dataset() is deprecated due to security risks. Users should temporarily use datasets<4.0.0 to be able to use our datasets. We are working on this issue to reformat our datasets and make them compliant to the new requirements.

• —Important Note: The current version of the PubChem dataset includes a new entry, PUBCHEM_SMILES, which is equivalent to isomeric SMILES and contains both stereochemical and isotopic information. This entry is set to replace both PUBCHEM_OPENEYE_CAN_SMILES and PUBCHEM_OPENEYE_ISO_SMILES in the future. For further details, please refer to the PubChem documentation.

Table of Contents

• —PubChem Dataset (version 04-18-2025)
• —Table of Contents
• —Dataset Description
• —Dataset Summary
• —Dataset Structure
• —Data Instances
• —Data Fields
• —Data Splits and Configurations
• —Dataset Creation
• —Curation Rationale
• —Source Data
• —Initial Data Collection and Normalization
• —Personal and Sensitive Information
• —Considerations for Using the Data
• —Social Impact of Dataset
• —Additional Information
• —Dataset Curators
• —Licensing Information
• —Citation Information
• —Contributions

Dataset Description

• —Homepage: https://pubchem.ncbi.nlm.nih.gov
• —Paper: https://doi.org/10.1093/nar/gkac956
• —Point of Contact: Sunghwan Kim
• —Point of Contact: Mohammad Mostafanejad
• —Point of Contact: MolSSI-AI Hub

Dataset Summary

PubChem is a popular chemical information resource that serves a wide range of use cases. It is an open chemistry database at the National Institutes of Health (NIH). "Open" means that you can put your scientific data in PubChem and that others may use it. Since the launch in 2004, PubChem has become a key chemical information resource for scientists, students, and the general public. Each month our website and programmatic services provide data to several million users worldwide.

PubChem mostly contains small molecules, but also larger molecules such as nucleotides, carbohydrates, lipids, peptides, and chemically-modified macromolecules. PubChem collects information on chemical structures, identifiers, chemical and physical properties, biological activities, patents, health, safety, toxicity data, and many others.

This version of the dataset was prepared for the project BERTology of Molecular Property Prediction, which investigates the effects of various factors on the pre-training and fine-tuning performance of BERT-based chemical language models.

Dataset Structure

Data Instances

An example of a data instance is as follows:

{
'PUBCHEM_COMPOUND_CID': '12',
'PUBCHEM_COMPOUND_CANONICALIZED': '1',
'PUBCHEM_CACTVS_COMPLEXITY': 104.0,
'PUBCHEM_CACTVS_HBOND_ACCEPTOR': 4,
'PUBCHEM_CACTVS_HBOND_DONOR': 4,
'PUBCHEM_CACTVS_ROTATABLE_BOND': 0,
'PUBCHEM_CACTVS_SUBSKEYS': 'AAADcYBgOAAAAAAAAAAAAAAAAAAAAAAAAAAwAAAAAAAAAAABAAAAGgAACAAACASAkAAwBoAAAgCAACBCAAACAAAgIAAAiAAGiIgJJyKCERKAcAElwBUJmAfAYAQAAQAACAAAQAACAAAQAACAAAAAAAAAAA==',
'PUBCHEM_IUPAC_OPENEYE_NAME': 'benzene-1,2,3,5-tetrol',
'PUBCHEM_IUPAC_CAS_NAME': 'benzene-1,2,3,5-tetrol',
'PUBCHEM_IUPAC_NAME_MARKUP': 'benzene-1,2,3,5-tetrol',
'PUBCHEM_IUPAC_NAME': 'benzene-1,2,3,5-tetrol',
'PUBCHEM_IUPAC_SYSTEMATIC_NAME': 'benzene-1,2,3,5-tetrol',
'PUBCHEM_IUPAC_TRADITIONAL_NAME': 'benzene-1,2,3,5-tetrol',
'PUBCHEM_IUPAC_INCHI': 'InChI=1S/C6H6O4/c7-3-1-4(8)6(10)5(9)2-3/h1-2,7-10H',
'PUBCHEM_IUPAC_INCHIKEY': 'RDJUHLUBPADHNP-UHFFFAOYSA-N',
'PUBCHEM_XLOGP3_AA': None,
'PUBCHEM_EXACT_MASS': 142.02660867,
'PUBCHEM_MOLECULAR_FORMULA': 'C6H6O4',
'PUBCHEM_MOLECULAR_WEIGHT': 142.11,
'PUBCHEM_SMILES': 'C1=C(C=C(C(=C1O)O)O)O',
'PUBCHEM_OPENEYE_CAN_SMILES': 'C1=C(C=C(C(=C1O)O)O)O',
'PUBCHEM_OPENEYE_ISO_SMILES': 'C1=C(C=C(C(=C1O)O)O)O',
'PUBCHEM_CACTVS_TPSA': 80.9,
'PUBCHEM_MONOISOTOPIC_WEIGHT': 142.02660867,
'PUBCHEM_TOTAL_CHARGE': 0,
'PUBCHEM_HEAVY_ATOM_COUNT': 10,
'PUBCHEM_ATOM_DEF_STEREO_COUNT': 0,
'PUBCHEM_ATOM_UDEF_STEREO_COUNT': 0,
'PUBCHEM_BOND_DEF_STEREO_COUNT': 0,
'PUBCHEM_BOND_UDEF_STEREO_COUNT': 0,
'PUBCHEM_ISOTOPIC_ATOM_COUNT': 0,
'PUBCHEM_COMPONENT_COUNT': 1,
'PUBCHEM_CACTVS_TAUTO_COUNT': 15,
'PUBCHEM_COORDINATE_TYPE': [1, 5, 255],
'PUBCHEM_BONDANNOTATIONS': [5,
6,
8,
...,
8],
'COORDS': [4.269000053405762,
2.0,
0.0,
...,
0.0],
'ATOMIC_INDICES': [1, 2, 3, ..., 16],
'ATOMIC_SYMBOLS': ['O',
'O',
'O',
...,
'H'],
'ATOMIC_NUMBERS': [8, 8, 8, ..., 1],
'ATOMIC_FORMAL_CHARGES': [0, 0, 0, ..., 0],
'BOND_ORDERS': [1,
5,
1,
...,
1],
'PUBCHEM_XLOGP3': '0.8',
'PUBCHEM_NONSTANDARDBOND': None,
'PUBCHEM_REFERENCE_STANDARDIZATION': None
}

Data Fields

Data Splits and Configurations

The dataset has only one train split and one configuration/subset:

• —pubchem-04-18-2025 (default)

Dataset Creation

Curation Rationale

The present version of PubChem dataset has been extracted from its original ftp repository, transformed into a dictionary and stored in the .json format.

Source Data

The link to the original PubChem dataset FTP repository can be found here

Initial Data Collection and Normalization

Other than the changes detailed in Sec. Curation Rationale, no data modification has been performed on the PubChem dataset.

Personal and Sensitive Information

The PubChem dataset does not involve any personal or sensitive information.

Considerations for Using the Data

Social Impact of Dataset

The PubChem dataset paves the way for applications in drug discovery and materials science, among others.

Additional Information

Dataset Curators

• —Sunghwan Kim, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA
• —Jie Chen, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA
• —Tiejun Cheng, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA
• —Asta Gindulyte, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA
• —Jia He, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA
• —Siqian He, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA
• —Qingliang Li, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA
• —Benjamin A Shoemaker, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA
• —Paul A Thiessen, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA
• —Bo Yu, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA
• —Leonid Zaslavsky, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA
• —Jian Zhang, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA
• —Evan E Bolton, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, 20894 USA

Licensing Information

Free Public Domain License

Citation Information

@article{Kim:2022:D1373,
    author = {Kim, Sunghwan and Chen, Jie and Cheng, Tiejun and Gindulyte, Asta and He, Jia and He, Siqian and Li, Qingliang and Shoemaker, Benjamin A and Thiessen, Paul A and Yu, Bo and Zaslavsky, Leonid and Zhang, Jian and Bolton, Evan E},
    title = "{PubChem 2023 update}",
    journal = {Nucleic Acids Research},
    volume = {51},
    pages = {D1373-D1380},
    year = {2022},
    doi = {10.1093/nar/gkac956}
}

@misc{mostafanejad:2026:bertology,
      title={BERTology of Molecular Property Prediction},
      author={Mohammad Mostafanejad and Paul Saxe and T. Daniel Crawford},
      year={2026},
      eprint={2603.13627},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2603.13627},
}

Contributions

• —Mohammad Mostafanejad, The Molecular Sciences Software Institute (MolSSI)