ChEMBL
EMBL's European Bioinformatics Institute (EMBL-EBI)
Millions of hand-curated measurements of how drug-like molecules hit their biological targets.
ChEMBL is a manually curated database of bioactive molecules with drug-like properties, extracting binding, functional, and ADMET measurements from the medicinal chemistry literature, patents, and deposited screens. Each activity links a compound, an assay, and a protein target. Drug discovery teams and machine learning researchers use it to train and benchmark activity models.
Access
- Cost
- FreeFree web interface, API, and full database dumps.
- License
- Open licenseCC BY-SA 3.0
- Sign-up
- No sign-up
- How you get it
- API and Bulk downloadPostgreSQL dump, SQLite, SDF, JSON
What's in it
- One row is
- One bioactivity measurement linking a compound, an assay, and a target
- Covers
- Not geographic
- Years
- Literature from the 1970s-present
- Updated
- Irregular
- Size
- Not stated
Quality
A Global Core Biodata Resource with careful manual curation and versioned releases.
- Activity values from different assays are not directly comparable
- Literature extraction favors published, often positive, results
Connects to
- PubChem Compound Database: Join by InChIKey for broader compound coverage
- Reaxys: Commercial alternative with reaction and property data for many of the same compounds
Questions it could answer
- Which protein targets have the most measured inhibitors?
- How has the typical potency of published kinase inhibitors changed over time?