LOTUS: Natural Product Occurrences and Metadata from Wikidata
by Adriano Rutz / University of Geneva
Available on 1 platform
Sign in to view source links and access this dataset
Description
The LOTUS Initiative provides a structured, manually validated collection of unique associations between chemical structures, biological organisms, and scientific references. Its metadata table includes structural descriptors like InChI and SMILES, chemical classifications from NPClassifier and ClassyFire, biological taxonomy from Open Tree of Life, and literature identifiers such as DOI and PMID. This dataset serves as a central resource for open research on natural products, linking chemistry, biology, and published evidence.
Use Cases
Identifying known natural products from a specific organism based on organism taxonomy.
Validating chemical-biological associations using the manual validation status flag.
Linking chemical structures to their published evidence via DOI and PMID references.
Classifying compounds using hierarchical chemical ontologies like ClassyFire and NPClassifier.
Enriching chemical databases with structural properties like molecular formula and exact mass.
Strengths
Data is manually validated for core associations between structures, organisms, and references.
Metadata is enriched from authoritative sources including PubChem, Open Tree of Life, and chemical classification systems.
Dataset is openly accessible under an Open Access (green) license.
Cross-platform presence on paperswithcode and Wikidata indicates community use and importance.
Limitations
Key quantitative details like row count and file size are not provided in any source.
The specific date of the last update is unknown, limiting assessment of freshness.
Column names are not explicitly listed, requiring inference from the file descriptions.
Provenance
Source
Wikidata, curated as part of the LOTUS Initiative (https://doi.org/10.7554/eLife.70780).
Collection Method
Exported from live Wikidata using the provided `lotus_exporter.py` script, generating frozen CSV files and a change report.
The dataset is described as 'frozen' exports from Wikidata, implying periodic snapshots rather than a live stream. The generator script is provided for reproducibility.