Activity Classes From Different Categories: 102 ChEMBL Classes with Fingerprints

A collection of 102 activity classes assembled from ChEMBL version 20, classified by difficulty for compound recall. The dataset also includes 10,000 randomly selected ZINC compounds. Swarit Jasial from the University of Bonn compiled this data, which provides MACCS and ECFP4 fingerprints for each compound.

Use Cases

• Benchmarking fingerprint performance for compound recall based on the provided activity class difficulty labels.
• Training classification models to predict compound activity based on MACCS and ECFP4 fingerprints.
• Analyzing the relationship between molecular structure and biological activity using the provided fingerprint data.
• Comparing the chemical space of known active compounds (from ChEMBL) against a random sample from the ZINC database.

Strengths

• Includes 102 distinct activity classes with expert-assigned difficulty labels ('easy', 'preferred/intermediate', 'difficult').
• Provides two standard molecular fingerprint types (MACCS and ECFP4) for each compound.
• Contains a reference set of 10,000 randomly selected compounds from the ZINC database.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the main activity class data is unknown, which may limit suitability assessment.
• Last update date is unknown; freshness unverified.

Provenance

Source

ChEMBL version 20 and the ZINC database.

Collection Method

Activity classes were assembled from ChEMBL and classified; ZINC compounds were randomly selected.

Time Range

Temporal coverage is not specified.

Freshness

Last updated date is unknown.

Geography

Spatial coverage is not specified.