ChEMBL 36: Bioactive Molecules with Drug-Like Properties

ChEMBL 36 contains over 2.4 million bioactive molecules from the manually curated ChEMBL database. The dataset was converted to HuggingFace format by user lukaskim and includes canonical SMILES representations. It was last updated on the platform in May 2026.

Use Cases

• Predict molecular properties based on SMILES strings.
• Train generative models for molecular design using canonical SMILES representations.
• Screen for potential drug candidates based on bioactivity data implied by the source database.
• Perform similarity searches across bioactive chemical space using the provided identifiers.

Strengths

• Contains over 2.4 million rows of bioactive compounds.
• Source data is from the ChEMBL database, which is manually curated by EMBL-EBI.
• Includes canonical SMILES and standard InChI representations for each molecule.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the specific HuggingFace instance is unknown, which may limit suitability assessment.
• Freshness should be verified as the source database is updated beyond this 2026 snapshot.

Provenance

Source

ChEMBL database (EMBL-EBI)

Collection Method

Manually curated database of bioactive molecules.

Freshness

Last updated 2026-05-05 21:57:42.