Molecular docking is a fast and efficient computational method for the prediction of the binding mode and binding affinity between a ligand and a target protein at the atomic level. However, the performance of current docking programs is less than satisfactory. Herein, with a focus on free programs and scoring functions, the performances of LeDock and three standalone scoring functions were tested by 195 high-quality protein–ligand complexes. Results showed that the success rate for the best pose of the free available docking program LeDock achieved 89.20%, indicative of a strong sampling power. Based on the poses generated by LeDock, a comparative evaluation on other three non-commercial scoring functions, including DSX (DrugScore X), PoseScore and X-score was performed. Among all the evaluated scoring functions, DSX and X-score exhibited the best scoring power and ranking power, respectively. The performances of LeDock, DSX and X-score were similar in docking power test, which was much better than the PoseScore. Accordingly, it was suggested that the combination of pose sampling by LeDock with rescoring by DSX or X-score could improve the prediction accuracy of molecular docking and applied in the lead discovery.

中文翻译：

---

评估提高分子对接性能的组合策略

分子对接是一种快速有效的计算方法，用于在原子水平上预测配体与靶蛋白之间的结合模式和结合亲和力。然而，目前的对接程序的性能却不尽如人意。在此，以免费程序和评分功能为重点，LeDock 的性能和三个独立评分功能通过 195 种高质量蛋白质-配体复合物进行了测试。结果表明，免费可用的对接程序LeDock的最佳姿势的成功率达到了89.20%，表明了强大的采样能力。基于 LeDock 生成的姿势，对其他三个非商业评分功能，包括 DSX (DrugScore X)、PoseScore 和 X-score 进行了比较评估。在所有评估的评分函数中，DSX 和 X-score 分别表现出最好的得分能力和排名能力。LeDock、DSX和X-score在对接功率测试中的表现相近，比PoseScore好很多。因此，建议将 LeDock 的姿态采样与 DSX 或 X-score 的重新评分相结合，可以提高分子对接的预测精度，并应用于先导发现。