liuganghuggingface/moldata-bde_db2

BDE-db2 and external bond-energy tests

Original author release, pinned to a repository commit. Raw structures and scientific labels are retained.

Processed copy of bde_db2 from the molecular datasets collection. Structures are canonicalized and validated; labels are never averaged or filled. Prefer the original source for citation.

Where the data comes from

• Original data: https://github.com/patonlab/bde-db2
• Original study: BDE-db2 and external bond-energy tests publication

Data quality

Grade: caution (rough; rules on the hub page).

• info: computed. Label origin from the source's tags.
• good: valid structures. Rows RDKit cannot parse (kept in quarantine).
• good: 2% undefined stereo. Molecules with at least one unassigned stereocenter (BACE: 71%).
• info: 578 stereoisomer sets. Groups of stereoisomers in the sampled structures; their labels can differ sharply.
• bad: R² ceiling ≈ 0.66. Replicate SD is 0.58× the label SD, so no model can explain more than about 66% of the variance.
• good: 100% label coverage. Median share of molecules labelled per task; missing labels are unmeasured, not negative.
• bad: 57 molecules per task. Median molecules with a label per task.
• good: source split. The source defines a split; shared structures between its parts leak information.

Tables

Config	Splits	Rows	Targets
000_bde_rdf_with_multi_halo_cfc_model_3	test, train, validation	834,066	bde (regression, not_verified), bdfe (regression, not_verified)
001_test_set_exp_bdes	full	82	exp_kcal_mol (regression, kcal/mol)
002_test_set_natural_products	full	81	bde (regression, not_verified), bdfe (regression, not_verified)
003_test_set_pfas	full	1,162	bde (regression, not_verified), bdfe (regression, not_verified)

data/ holds parseable rows with identical rows collapsed. observations/ holds every source row, including unparseable structures, for auditing. Added columns: SMILES_original, SMILES_canonical (RDKit; stereo, salts, and * kept), structure_key (JSON list of canonical components; use it to deduplicate or group), structure_status, wildcard_atom_count, source_row_first, observation_count, source_dataset, source_table. All original columns are strings, so censored values such as <10 survive; cast labels yourself.

Load

from torch_molecule.datasets import load_dataset
data = load_dataset("bde_db2")

from datasets import load_dataset
ds = load_dataset("liuganghuggingface/moldata-bde_db2", "000_bde_rdf_with_multi_halo_cfc_model_3")

Known issues

• Training values are computed; the separately identified experimental external test must not be mixed into a DFT-only label pool.
• Bond indices and radical fragments are part of task identity. Repeated molecules with different bonds are distinct observations.
• Model-3 expanded training table selected; original model-1 and model-2 overlapping training tables are not separately copied.
• Acquisition failed for LICENSE: 404 Client Error: Not Found for url: https://raw.githubusercontent.com/patonlab/bde-db2/be02a7396b93a3228efbd1363a00b8a5f1151adc/LICENSE

License

Not stated by the source. Redistributed with the collection maintainer's approval; check the original source (https://github.com/patonlab/bde-db2) for its terms before reuse.

Citation

Cite the original study and data source:

• BDE-db2 and external bond-energy tests publication