ChEMBL datasets collection
Automated retrieval and preprocessing of ChEMBL bioactivity data for QSAR modeling.
Scientific problem
ChEMBL contains heterogeneous assay annotations and activity records that require consistent filtering and preprocessing before QSAR modeling.
Motivation
The pipeline was built to make repeated extraction of task-specific bioactivity datasets faster and reproducible.
My role
- Developed the data-retrieval and preprocessing pipeline.
- Implemented filtering for specific activity modes, including agonists, antagonists, and inverse agonists.
Methodology
- Programmatic extraction of ChEMBL bioactivity records.
- Activity-type filtering, structure preprocessing, and dataset quality control.
- Automatic dataset standardization for downstream QSAR modeling.
Results
- Produced reusable, reproducible datasets for ligand-based modeling tasks.
Software
- Python
- RDKit
- pandas
- SQL
- API