论文部分内容阅读
以35种嘌呤骨架类热休克蛋白90(Hsp90)化合物为研究对象,以文献[3]的8个变量构成自变量集,提出1种改进的MC GEP算法对该类化合物抗癌活性pEC_(50)做定量结构活性关系研究。按文献[3]对所有35种嘌呤类化合物建模,本文GEP模型与文献[3]的回归模型计算结果决定系数R~2分别为0.821 4和0.738 0。进一步用k均值聚类算法将此35种化合物划分为训练集和预测集,分别采用改进的MC GEP算法、v-SVM和ε-SVM算法基于训练集建模,本文建立的GEP模型训练和预测结果R~2分别为0.808 0和0.745 5,而v-SVM和ε-SVM模型对预测集的预测结果R~2分别为0.204 6和0.410 3,均低于0.5。研究表明,本文提出的改进MC GEP算法函数发现能力较强,建立的QSAR模型预测性能好。
A total of 35 purine-type Hsp90 compounds were selected as the research objects. Eight independent variable sets were constructed in [3], and an improved MC GEP algorithm was proposed to determine the anticancer activity of these compounds. ) To do quantitative structure activity relationship research. According to the literature [3], all 35 purine compounds were modeled. The regression coefficients of GEP model and literature [3] determined that the coefficients R ~ 2 were 0.821 4 and 0.738 0 respectively. The 35 compounds were further divided into training set and prediction set by k-means clustering algorithm. The improved MC GEP algorithm, v-SVM and ε-SVM algorithm were respectively used to train the training set and prediction set based on training set model. The results of R ~ 2 were 0.808 0 and 0.745 5, respectively. The predicted results of prediction set V-SVM and ε-SVM for R 2 were 0.204 6 and 0.410 3, both of which were lower than 0.5. The research shows that the improved MC GEP algorithm proposed in this paper has strong ability of finding functions and the established QSAR model has good prediction performance.