MUV is a benchmark dataset used to assess the performance of ligand-based virtual screening methods. It contains challenging targets with decoys chosen to be chemically similar to active compounds, making it a valuable benchmark for similarity-based approaches

The QED dataset contains molecules with drug-likeness scores, which represent the similarity of compounds to known drugs. It is used to evaluate AI models' ability to prioritize compounds with drug-like properties

Similar to TPC-H, TPC-DS is another benchmark by the Transaction Processing Performance Council that focuses on decision support systems but provides a more complex and realistic workload.

An AI benchmark for molecular machine learning. It provides benchmark datasets and evaluation metrics for various drug discovery tasks

QM7 and QM9 are benchmark datasets used for evaluating AI models in quantum chemistry tasks. QM7 contains quantum mechanical calculations of energies, enthalpies, and dipole moments of small molecules, while QM9 includes additional properties like atomization energies and HOMO-LUMO gaps

The Transaction Processing Performance Council's TPC-H benchmark measures the performance of decision support systems by executing a set of ad-hoc queries on large volumes of data.

PDBbind is a benchmark dataset for protein-ligand binding affinity prediction. It includes experimentally determined binding affinities for protein-ligand complexes and is used to evaluate the accuracy of AI models in predicting binding affinities

SIDER (Side Effect Resource) is a database that provides information on marketed drugs and their side effects. It is used to evaluate AI models' performance in predicting potential adverse drug reactions and toxicity

The Tox21 dataset includes biological activity data for thousands of compounds across various cell-based assays. It is commonly used as a benchmark for toxicity prediction models.

The PubChem BioAssay Database is a repository of biological assay data, including screening results from high-throughput screening experiments. It provides a diverse set of bioactivity data that can be used to assess AI models' performance in predicting compound activities

This benchmark evaluates the performance of online transaction processing (OLTP) systems by simulating a complete order-entry and inventory control system.

YCSB is a benchmark that evaluates the performance of key-value and cloud storage systems, simulating workloads of different applications.