QSAR Benchmark Datasets for Molecular Activity Prediction

This collection comprises datasets from the QSAR-TID series, each pairing chemical compound features with associated biological activity endpoints. The data is standardized and hosted on the OpenML platform, facilitating direct use in machine learning workflows. Together, these datasets serve as a comprehensive benchmark suite for developing and evaluating Quantitative Structure-Activity Relationship models. Researchers can use this collection to test predictive algorithms across a variety of molecular targets and activity types.