To the best of our knowledge, we provide here for the first time several high throughput screening data mimicking real life scenarios in which all true positives and true negatives are known. Such benchmarking data are, to our opinion, much more valuable than commonly used benchmarks in which actives (usually high affinity ligands) are mixed with chemically similar decoys of unknown affinity for the intended target.（J. Chem. Inf. Model. 2019, 59, 573-585）

即测试一个虚拟筛选方法的优劣时，使用具有客观测试数据的PubChem数据集可能比使用设置decoys的DUD-E数据集可能要更优。