Datasetpublic

ThorKl/theobroma

THEOBROMA v1.36 An aggregated open database of 1,132,805 natural products from 29 sources, with per-compound license auditing, three-tier classification provenance, and stereochemistry-aware deduplication. Live instance: https://theobroma.l3s.uni-hannover.de Archival record: https://doi.org/10.5281/zenodo.20443051 (concept DOI, resolves to latest) This release: https://doi.org/10.5281/zenodo.22816330 Preprint: https://doi.org/10.64898/2026.06.12.731585 Licensing… See the full description on the dataset page: https://huggingface.co/datasets/ThorKl/theobroma.

sourceHugging Faceotherupdated 4d agoView on Hugging Face

0likes180downloads

Dataset Card

THEOBROMA v1.36

An aggregated open database of 1,132,805 natural products from 29 sources, with per-compound license auditing, three-tier classification provenance, and stereochemistry-aware deduplication.

—Live instance: https://theobroma.l3s.uni-hannover.de
—Archival record: https://doi.org/10.5281/zenodo.20443051 (concept DOI, resolves to latest)
—This release: https://doi.org/10.5281/zenodo.22816330
—Preprint: https://doi.org/10.64898/2026.06.12.731585

Licensing

This dataset is not uniformly licensed. Each compound carries a resolved license tier in the license_tier column, computed under a most-restrictive-wins rule across all attesting sources.

Tier	Compounds
CC BY 4.0	887,999
CC BY-NC 4.0	229,397
CC0	8,243
Unspecified	7,166

896,242 compounds (79.12%) are open-licensed under the most restrictive resolution. A least-restrictive bound is also recorded in tier_rank_min, under which 1,029,587 compounds (90.9%) reach CC BY or better from at least one attesting source. Downstream use must be governed by license_tier per compound, not by any single license applied to the whole dataset. The license_attestations table records which source asserted which terms for each compound, and source_licenses maps each source to its tier.

Tables

File	Rows	Contents
`compounds`	1,132,805	structures, identifiers, descriptors, classification, license tier
`compound_taxonomy`	1,504,425	per-source organism attestations
`resolved_taxonomy`	1,132,805	majority-voted lineage
`license_attestations`	1,512,594	per-source license attestation chain
`source_licenses`	53	source-to-tier map; `in_corpus` marks the 29 integrated sources
`synonyms`	481,817	compound synonyms
`regions`	772,870	macro-region mapping
`admet`	1,132,789	47-column in silico ADMET surface
`scaffolds`	1,132,805	Bemis-Murcko scaffolds

Note that license_tier appears in two tables with different meanings: in compounds it is the resolved tier after most-restrictive-wins across all attesting sources, and in license_attestations it is the tier stated by that individual source. Use suffixes when joining the two.

Nested columns are stored as Arrow structs rather than JSON strings: chebi_xrefs, the three lineage columns in compound_taxonomy, and native_distribution. secondary_kingdoms in resolved_taxonomy is a list of strings.