SQuIRL Dataset

SQuIRL is a dataset of 133883 small molecules from the QM9 dataset (doi: 10.1038/sdata.2014.22), with the following properties:SMILES stringsMolecular geometries (Cartesian) (Å)Functional groupsPrincipal moments of inertia (amu x Å2)Force constants (normal modes) (mDyne/Å (= 0.1 N/m))Harmonic IR intensities (km x mol−1)Harmonic vibrational wavenumbers (cm−1)Point group symmetryDipole moment (Debye)Dipole polarizability (a.u. (bohr3))Quadrupole moment (Debye x Å)Electronic energy (SCF) (Hartree)Zero-point vibrational energy (ZPVE) (Hartree)Thermal correction to energy (Hartree)Thermal correction to enthalpy (Hartree)Thermal correction to Gibbs free energy (Hartree)Sum of electronic and ZPVE (Hartree)Sum of electronic and thermal energies (Hartree)Sum of electronic and thermal enthalpies (Hartree)Sum of electronic and thermal free energies (Hartree)Heat capacity CV (thermal analysis) (cal x mol−1 x K−1)Warnings / notesAtomic numbersAll calculated at ωB97X-D/aug-cc-pVTZ level of theory from original QM9 structures.The records are stored in an HDF5 format with the following structure:data/├── molecule_ID/│ ├──▶ warnings [str]│ ├── electrostatics/│ │ ├── dipole_moment [shape=(3,), dtype=float64]│ │ ├── dipole_polarizability [shape=(3, 3), dtype=float64]│ │ ├── energy [dtype=float64]│ │ └── quadrupole_moment [shape=(3, 3), dtype=float64]│ ├── infrared/│ │ ├── force_constants [shape=(n_peaks,), dtype=float64]│ │ ├── magnitudes [shape=(n_peaks,), dtype=float64]│ │ ├── symmetries [shape=(n_peaks,), dtype=object]│ │ └── wavenumbers [shape=(n_peaks,), dtype=float64]│ ├── structure/│ │ ├── pos [shape=(n_atoms, 3), dtype=float64]│ │ ├── principal_moments [shape=(3,), dtype=float64]│ │ ├── z [shape=(n_atoms,), dtype=int64]│ │ ├──▶ smiles [str]│ │ └──▶ functional_groups [str]│ ├── thermochemistry/│ │ ├── electronic_energy_sum [dtype=float64]│ │ ├── electronic_enthalpy_sum [dtype=float64]│ │ ├── electronic_free_energy_sum [dtype=float64]│ │ ├── electronic_zero_point_sum [dtype=float64]│ │ ├── entropy [dtype=float64]│ │ ├── heat_capacity [dtype=float64]│ │ ├── thermal_energy_corr [dtype=float64]│ │ ├── thermal_enthalpy_corr [dtype=float64]│ │ ├── thermal_gibbs_corr [dtype=float64]│ │ ├── total_thermal_energy [dtype=float64]│ │ └── zero_point_corr [dtype=float64]│ │. .. .. .We provide an an example Jupyter Notebook that can be used to explore the dataset and understand how the records are accessed. This can be used after running the following commands in a terminal with an active python3 environment:python3 pip install jupyterjupyter notebook SQuIRL_Explorer.ipynbIn addition, a companion CSV file is also available to enable filtering, querying, and exploratory analysis. It contains key per-molecule attributes including:internal HDF5 path, SMILES representation, functional group annotations, warning flags, number of atoms, number of vibrational peaks, dipole components and magnitude, principal moments of inertia, electronic and thermochemical quantities.