Molecule Bioactivity Data for Four Protein Targets

8,711 molecules with measured binding activity against four distinct protein targets. The dataset originates from the ChEMBL database, a major public repository for bioactive molecules. The specific creation date and author are not provided.

Use Cases

• Train multi-label classification models to predict which of the four protein targets a molecule binds to based on its structural features.
• Build regression models to predict binding affinity values for each target, using molecular descriptors.
• Analyze structural patterns and chemical scaffolds common to molecules active against specific target combinations.
• Use the dataset as a benchmark for evaluating virtual screening or multi-task learning algorithms in cheminformatics.

Strengths

• Contains 8,711 molecules, providing a substantial sample for model training.
• Focuses on four distinct protein targets, enabling multi-target analysis.

Limitations

• The specific bioactivity measurements, molecular features, and target identities are not detailed, limiting immediate analysis.
• The dataset size is moderate; models may benefit from additional data for complex multi-target predictions.

Provenance

Source

ChEMBL database

Collection Method

Curated extraction of molecules with confirmed binding activity to four specific targets.