Division of Applied Life Science (BK21 Program), Environmental Biotechnology National Core Research Center (EB-NCRC), Gyeongsang National University, Jinju 660-701, Republic of Korea.
Eur J Med Chem. 2010 Jun;45(6):2132-40. doi: 10.1016/j.ejmech.2010.01.016. Epub 2010 Feb 4.
Chemical features based 3D pharmacophore models were developed for HSP90 based on the known inhibitors using Discovery Studio V2.1. An optimal pharmacophore model was brought forth and validated using a decoy set, external test set and Fischer's randomization method. The best five features pharmacophore model, Hypo1, includes two hydrogen bond acceptors, three hydrophobic features, which has the highest correlation coefficient (0.93), cost difference (73.88), low RMS (1.24), as well as it shows a high goodness of fit and enrichment factor. Hypo1 was used as a 3D query for virtual screening to retrieve potential inhibitors from Maybridge and Scaffold databases. The hit compounds were subsequently subjected to molecular docking studies and finally, 36 compounds were obtained based on consensus scoring function.
基于已知抑制剂，使用 Discovery Studio V2.1 为 HSP90 开发了基于化学特征的 3D 药效团模型。使用诱饵集、外部测试集和 Fischer 随机化方法对最优药效团模型进行了验证。最佳的五个特征药效团模型 Hypo1 包括两个氢键受体和三个疏水性特征，具有最高的相关系数（0.93）、成本差异（73.88）、低 RMS（1.24），并且表现出良好的拟合度和富集因子。Hypo1 被用作虚拟筛选的 3D 查询，以从 Maybridge 和 Scaffold 数据库中检索潜在的抑制剂。随后对命中化合物进行分子对接研究，最后根据共识评分函数获得 36 个化合物。
