摘要
脑部淀粉样β多肽(β-amyloid peptide,Aβ)的纤维化沉积是阿尔兹海默症(Alzheimer's disease,AD)主要病理特征之一。因此,开发Aβ聚集抑制剂对AD治疗具有重要意义。本文采用支持向量机(Support Vector Machine,SVM)方法分别构建Aβ和Aβ_(40)聚集抑制活性分类模型。采用五重交叉验证筛选模型参数,并通过留一法验证模型。结果表明,Aβ和Aβ_(40)聚集抑制剂分类模型对训练集的预测精度分别为98.4%和93.2%,对测试集的预测精度分别为73.3%和75.0%。我们利用预测效果较好的Aβ_(40)聚集抑制剂分类模型从中药化学数据库(Traditional Chinese Medicines Database,TCMD)中筛选出17种具有潜在Aβ抑制活性的中药化合物。经统计分析,含有以上命中化台物最多的中草药植物分别为雷公藤(Tripterygium wilfordii)和猴头菌(Hericium erinaceus)。这为从中药化合物中发现新的AD治疗药物提供了理论指导。
Aggregated β-amyloid peptide （Aβ） is assumed to be one of the most relevant neuropathological markers for Alzheimer＇s disease （AD）. Therefore, inhibitors of Aβ aggregation may play an important therapeutic role in treatment of AD. In the present study, Support Vector Machine （SVM） was employed to establish the classification and prediction models of Aβ and Aβ40 aggregation inhibitors. The 5-fold cross validation was employed to identify the parameters of the SVM in the model training process and Leave-one-out cross validation （LOOCV） was used to evaluate the predictive power of models. The overall prediction accuracy of the established models for the training set are 98.4% and 93.2%, and that for test set are 73.3% and 75.0% respectively, which showed that our models have good prediction abilities. Subsequently, the classification model of Aβ40 aggregation inhibitors was used to screen Traditional Chinese Medicine Database （TCMD）. Finally, a total of 17 compounds were identified as potential Aβ inhibitors, which are mostly found in Chinese Herbs Tripterygium wilfordii and Hericium erinaceus. The findings may provide hints for the treatment of AD.
出处
《计算机与应用化学》
CAS
CSCD
北大核心
2013年第2期133-136,共4页
Computers and Applied Chemistry
基金
国家自然科学基金资助项目(31201952)
中国博士后科学基金资助项目(20110491166)
