SARS-CoV-2蛋白质-配体对接替代模型数据集

本数据集由阿贡国家实验室数据科学与学习部开发，包含2亿个3D复合结构和2D结构评分，涵盖了SARS-CoV-2蛋白质组的15个受体或结合位点上的1300万个“现货”分子。数据集旨在通过深度学习加速虚拟筛选，提高对接模型的吞吐量，比传统对接协议快六个数量级。该数据集的应用领域主要集中在药物发现，特别是针对SARS-CoV-2的药物筛选，以期快速找到有效的药物候选分子。

This dataset was developed by the Data Science and Learning Division of Argonne National Laboratory. It contains 200 million 3D complex structures and their corresponding 2D structure scores, covering 13 million "stock" (commercially available) molecules targeting 15 receptors or binding sites on the SARS-CoV-2 proteome. This dataset is designed to accelerate virtual screening via deep learning, boosting the throughput of docking models by six orders of magnitude compared to traditional docking protocols. Its primary application areas lie in drug discovery, especially SARS-CoV-2-targeted drug screening, with the aim of rapidly identifying effective drug candidate molecules.