State Key Laboratory of Biotherapy and Cancer Center, West China Hospital, West China Medical School, Sichuan University, Sichuan 610041, China.
J Mol Graph Model. 2013 Jul;44:278-85. doi: 10.1016/j.jmgm.2013.07.005. Epub 2013 Jul 23.
Herein, a combined molecular docking-based and pharmacophore-based target prediction strategy is presented, in which a probabilistic fusion method is suggested for target ranking. Establishment and validation of the combined strategy are described. A target database, termed TargetDB, was firstly constructed, which contains 1105 drug targets. Based on TargetDB, the molecular docking-based target prediction and pharmacophore-based target prediction protocols were established. A probabilistic fusion method was then developed by constructing probability assignment curves (PACs) against a set of selected targets. Finally the workflow for the combined molecular docking-based and pharmacophore-based target prediction strategy was established. Evaluations of the performance of the combined strategy were carried out against a set of structurally different single-target compounds and a well-known multi-target drug, 4H-tamoxifen, which results showed that the combined strategy consistently outperformed the sole use of docking-based and pharmacophore-based methods. Overall, this investigation provides a possible way for improving the accuracy of in silico target prediction and a method for target ranking.
在此，提出了一种基于分子对接和基于药效团的组合靶标预测策略，其中建议使用概率融合方法进行靶标排序。描述了组合策略的建立和验证。首先构建了一个称为 TargetDB 的靶标数据库，其中包含 1105 个药物靶标。基于 TargetDB，建立了基于分子对接的靶标预测和基于药效团的靶标预测方案。然后通过构建针对一组选定靶标的概率分配曲线 (PAC) 开发了一种概率融合方法。最后建立了基于分子对接和基于药效团的组合靶标预测策略的工作流程。针对一组结构不同的单靶化合物和一种著名的多靶药物 4H-他莫昔芬对组合策略的性能进行了评估，结果表明组合策略始终优于单独使用基于对接和基于药效团的方法。总的来说，这项研究为提高计算靶标预测的准确性提供了一种可能的方法，也为靶标排序提供了一种方法。
