ROC曲线分析在评价计算机虚拟筛选乙酰胆碱酯酶抑制剂中的应用  

Application of ROC Curve Analysis in Evaluating the Virtual Screening of ACHE Inhibitors

作　　者：葛庆波[1] 罗巧云[1] 王真真[1] 方兆庚 刘述[2] 

机构地区：[1]解放军第458医院,广州510602 [2]中山大学中山医学院解剖学教研室海洋微生物功能分子广东省高校重点实验室,广州510080

出　　处：《中国医药导刊》2010年第10期1746-1748,共3页Chinese Journal of Medicinal Guide

摘　　要：目的:将ROC曲线分析应用在计算机虚拟药物筛选研究领域中。方法:利用分子对接软件以乙酰胆碱酯酶的三维晶体结构为靶,虚拟筛选40个抑制剂和3960个无活性化合物,利用5种打分函数(Surflex_score,D_score,G_score,PMF_score,ChemScore)对筛选结果打分排序,应用ROC曲线分析打分排序后的结果。结果:ROC曲线分析出利用ChemScore打分函数得出的曲线下的面积(AUC)最大,为0.91;如果特异度作为优先考虑因素时,应用ChemScore打分函数可以获得最多的活性化合物;如果灵敏度作为优先考虑因素时,应用ChemScore打分函数也可以获得最多的活性化合物。结论:ROC曲线分析同时考虑到了灵敏度和特异度,在计算机虚拟药物筛选过程中,可以根据实际情况设定打分函数的闽值,获得尽可能多的活性化合物。Objective： Application of ROC curve analysis in evaluating the virtual screening of AChE inhibitors.Materials：Virtual screen for ACHE inhibitors using different scoring functions（Surflex score,D score,G score,PMF score,ChemScore）,the screening library was composed of 40 ACHE inhibitors and 3960 inactive compounds. Five scoring functions were used for ranking,and application of ROC curve analysis in evaluating the results of ranking.Resnlts：Using ChemScore to rank the database, the area under the curve （AUC） was 0.91 .If we prefered specificity,more active compound were obtained by ChemScore;if we preferred sensitivity, more active compound s were obtained by ChemScore.Conclusion： ROC curve consists of sensitivity and specificity together,we can use it to set the right threshold after virtual screening to obtain active compounds as many as possible.

关 键 词：ROC曲线 乙酰胆碱酯酶 虚拟筛选 打分函数 

分 类 号：R927[医药卫生—药学]