molssiai-hub/pubchemqc-pm6

PubChemQC PM6 Dataset

Dataset Summary

The PubChemQC PM6 dataset consists of optimized molecular geometries and electronic properties calculated by the PM6 method for 94.0% of the 91.6 million molecules cataloged in PubChem Compounds retrieved on August 29, 2016. In addition to neutral states, the cationic, anionic, and spin flipped electronic states of respectively 56.2%, 49.7%, and 41.3% of the molecules have also been studied. As such, the grand total of the PM6 calculations amounted to 221 million.

Dataset Structure

Data Instances

An example of a data instance is as follows:

{
  "cid": 1,
  "state": "S0",
  "pubchem-inchi": "InChI=1S/C9H17NO4/c1-7(11)14-8(5-9(12)13)6-10(2,3)4/h8H,5-6H2,1-4H3",
  "pubchem-charge": 0,
  "pubchem-version": "20160829",
  "name": "000000001.PM6.S0",
  "coordinates": [4.543146, -2.8411939999999998, -1.641861, ..., 4.345637],
  "atomic-numbers": [6, 6, 8, ..., 1],
  "atom-count": 31,
  "heavy-atom-count": 14,
  "core-electrons": [0, 0, 0,..., 0],
  "version": "2.0",
  "obabel-inchi": "InChI=1S/C9H17NO4/c1-7(11)14-8(5-9(12)13)6-10(2,3)4/h8H,5-6H2,1-4H3/t8-/m0/s1",
  "pm6-obabel-canonical-smiles": "[O]C(=O)C[C@@H](C[N](C)(C)C)OC(=O)C",
  "charge": 0,
  "energy-beta-gap": 8.458931156643,
  "energy-beta-homo": -8.8325434733795,
  "energy-beta-lumo": -0.37361231673649903,
  "energy-alpha-gap": 8.458931156643,
  "energy-alpha-homo": -8.8325434733795,
  "energy-alpha-lumo": -0.37361231673649903,
  "total-energy": -7.558832783207512,
  "enthalpy": -0.021744,
  "homos": [40],
  "orbital-energies": [
    [-34.149199782348006,
     -32.51760513475,
     -30.809274381311,
     ...,
     6.947610830966
    ]
  ],
  "mo-count": 73,
  "basis-count": 73,
  "temperature": 298.15,
  "multiplicity": 1,
  "number-of-atoms": 31,
  "mulliken-partial-charges": [-0.5888869999999999,
  0.6862539999999999,
  -0.580137,
  ...,
  0.18249
  ],
  "dipole-moment": 13.22226250268841,
  "frequencies": [26.1465, 37.9831, 50.1099, ..., 2779.2147],
  "intensities": [0.771, 11.6909, 2.5082, ..., 111.8945],
  "pubchem-multiplicity": 1,
  "pubchem-obabel-canonical-smiles": "[O-]C(=O)CC(C[N+](C)(C)C)OC(=O)C",
  "pubchem-isomeric-smiles": "CC(=O)OC(CC(=O)[O-])C[N+](C)(C)C",
  "pubchem-molecular-weight": 203.23558,
  "pubchem-molecular-formula": "C9H17NO4"
}

Data Fields

Field	Description
cid	Pubchem Compound ID
state	Electronic state
pubchem-inchi	InChI extracted from PubChem Compound entry
pubchem-charge	Molecular charge extracted from PubChem Compound entry
pubchem-version	PubChem Compound Database version
name	Name of the input file used for the PM6 calculation
coordinates	Cartesian coordinates of the molecular geometry optimized with PM6 method in Angstroem
atomic-numbers	An array of atomic numbers
atom-count	Number of atoms in the molecule
heavy-atom-count	Number of heavy atoms in the molecule
core-electrons	The number of core electrons in each atom's pseudopotentials
version	Version number
obabel-inchi	InChI of the structure generated by Open Babel
pm6-obabel-canonical-smiles	Canonical SMILES for the structure generated by Open Babel
charge	Molecular charge
energy-beta-gap	HOMO-LUMO energy gap for beta spin orbitals
energy-beta-homo	Energy of the highest-occupied molecular orbital (HOMO) of beta spin symmetry
energy-beta-lumo	Energy of the lowest-unoccupied molecular orbital (LUMO) of beta spin symmetry
energy-alpha-gap	HOMO-LUMO energy gap for alpha spin orbitals
energy-alpha-homo	Energy of the highest-occupied molecular orbital (HOMO) of alpha spin symmetry
energy-alpha-lumo	Energy of the lowest-unoccupied molecular orbital (LUMO) of alpha spin symmetry
total-energy	Total electronic energy of the molecule calculated using PM6 method
enthalpy	Enthalpy
homos	1D index array of the highest occupied molecular orbital (HOMOs) with one (two) element(s) for the (un)restricted wavefunctions
orbital-energies	1D array of orbital energies in hartree with one (two) member(s) for the (un)restricted wavefunction
mo-count	Number of molecular orbitals
basis-count	Number of basis functions
temperature	Temperature (in K)
multiplicity	Spin multiplicity
number-of-atoms	Number of atoms in the molecule
mulliken-partial-charges	Mulliken partial atomic charges
dipole-moment	Dipole moment
frequencies	Infrared (IR) frequencies (in cm$^{-1}$)
intensities	Infrared intensities
pubchem-multiplicity	Spin multiplicity of the molecule extracted from PubChem Compound
pubchem-obabel-canonical-smiles	Canonical SMILES of the molecule extracted from PubChem Compound generated by Open Babel
pubchem-isomeric-smiles	Isomeric SMILES of the molecule extracted from PubChem Compound calculated by the OpenEye's OEChem Toolkit
pubchem-molecular-weight	Molecular weight extracted from the PubChem Compound entry
pubchem-molecular-formula	Molecular formula extracted from the PubChem Compound entry

Data Splits and Configurations

The dataset has only one train split. The PubChemQC PM6 dataset has seven configurations/subsets:

• pm6opt (default)
• pm6opt_chon300nosalt
• pm6opt_chon500nosalt
• pm6opt_chnops500nosalt
• pm6opt_chnopsfcl300nosalt
• pm6opt_chnopsfcl500nosalt
• pm6opt_chnopsfclnakmgca500

How to Use the Dataset

Prerequisites

We recommend isolating your work in a virtualenv or conda environment. You can create a new conda environment, pubchemqc,

  conda create -n pubchemqc python=3.12

and activate it using the following command

  conda activate pubchemqc

Once the conda environment is activated, you can install the dependencies in it as shown below

  pip install huggingface_hub ijson

Accessing the Data

Once the required packages are installed, you can run the following code to access the data

  # import the modules
  from datasets import load_dataset

  # load the dataset with streaming
  hub_ds = load_dataset(path="molssiai-hub/pubchemqc-pm6",
                        name="pm6opt",
                        split="train",
                        streaming=True,
                        cache_dir="./tmp",
                        trust_remote_code=True)

  # fetch a batch of 32 samples from the dataset
  ds = list(hub_ds.take(32))

The argument name by default is set to pm6opt which refers to the entire dataset. Other configurations (subsets), listed in Sec. Data Splits and Configurations, can also be selected.

The split must be set to train as it is the only split in our dataset. We recommend using streaming=True to avoid downloading the entire dataset on disk. The cache_dir allows us to store the Hugging Face datasets' and models' artifacts in a non-default directory (by default, it is set to ~/.cache/huggingface). As we are using a custom load script, the trust_remote_code argument should also be set to True.

Dataset Creation

Curation Rationale

The present version of PubChemQC PM6 dataset has been extracted from its original Postgresql database, transformed into a dictionary and stored in the .json format.

Source Data

The link to the original PubChemQC PM6 dataset 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 PubChemQC PM6 dataset.

Personal and Sensitive Information

The PubChemQC PM6 dataset does not involve any personal or sensitive information.

Considerations for Using the Data

Social Impact of Dataset

The PubChemQC PM6 dataset paves the way for applications in drug discovery and materials science, among others.

Additional Information

Dataset Curators

• Maho Nakata, RIKEN Cluster for Pioneering Research, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan
• Tomomi Shimazaki, Graduate School of System Informatics, Kobe University, 1-1 Rokkodai-cho, Nada-ku, Kobe, Hyogo 657-8501, JAPAN
• Masatomo Hashimoto, Software Technology and Artificial Intelligence Research Laboratory, Chiba Institute of Technology, 2-17-1 Tsudanuma, Narashino, Chiba 275-0016, Japan
• Toshiyuki Maeda, Software Technology and Artificial Intelligence Research Laboratory, Chiba Institute of Technology, 2-17-1 Tsudanuma, Narashino, Chiba 275-0016, Japan

Licensing Information

Creative Commons Attribution 4.0 International License

Citation Information

@article{Nakata:2020:5891,
   author = {Maho Nakata and Tomomi Shimazaki and Masatomo Hashimoto and Toshiyuki Maeda},
   doi = {10.1021/acs.jcim.0c00740},
   issn = {1549960X},
   issue = {12},
   journal = {Journal of Chemical Information and Modeling},W
   month = {12},
   pages = {5891-5899},
   pmid = {33104339},
   publisher = {American Chemical Society},
   title = {PubChemQC PM6: Data Sets of 221 Million Molecules with Optimized Molecular Geometries and Electronic Properties},
   volume = {60},
   url = {https://pubs.acs.org/doi/abs/10.1021/acs.jcim.0c00740},
   year = {2020},
}

Contributions

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