中文摘要：

以作用于鼠肝脏细胞的21个3-羟基-3-甲基戊二酰辅酶A（HMG-CoA）还原酶抑制剂（RI）为训练集，训练集化合物具备结构多样性，来源于相同药理模型，活性值IC50范围在0．3—8000nmol·L^-1．利用Catalyst计算HMG-CoA还原酶抑制剂最优药效团由一个氢键受体，一个氢键给体，一个疏水基团和一个芳香环特征组成．药效团模型Fixed cost值，Total cost值和Configuration cost值分别为88．75、111．5和16．98．训练集化合物活性计算值与实测值相关系数为0．8883，偏差值为1．269，交叉验证结果表明，药效团模型具有较高的置信度，对测试集化合物活性值的预测结果显示有较好的预测能力，可用于数据库搜索发现新的具有该活性的化合物，也可用于中药或天然产物药物的研究开发．

英文摘要：

A three-dimensional pharmacophore model of 3-hydro-3-methyl glutaryl coenzyme A （HMG-CoA） reductase inhibitors （RI） was developed based on 21 HMG-CoA reductase inhibitors from a rat liver. The selected training set had great diversity in both molecular structure and bioactivity as required by HypoGen program in the Catalyst software. The inhibitors in training set showed HMG-CoA RI inhibiting activity with IC50 values of 0.3-8000 nmol ·L^-1. The best statistical hypothesis, consisting of four features, one hydrogen-bond acceptor, one hydrogen-bond donor, one ring aromatic feature, and one hydrophobic point, had a correlation coefficient of 0.8883, a root-mean-square （RMS） deviation of 1.269, a Fixed cost of 88.75, a Total cost of 111.5, and a Configuration cost of 16.98, this hypothesis had highly predictive ability. The predictive ability was approved by the results of the activity estimated by mapping the compounds of the testing set with it. The cross-validation provided strong confidence on this hypothesis. This pharmacophore model can contribute to the finding and designing of new-type HMG-CoA reductase inhibitors, and to the development of traditional Chinese medicine and materia medica.