基于混合选择的GEP算法及其在喹诺酮羧酸类HIV-1整合酶抑制剂活性的QSAR研究中的应用  

QSAR study on quinolone carboxylic acids as HIV-1 integrase inhibitors based on GEPMS algorithm

作　　者：张运陶[1] 付伟忠[1] 

机构地区：[1]西华师范大学应用化学研究所,四川南充637002

出　　处：《计算机与应用化学》2011年第1期83-87,共5页Computers and Applied Chemistry

摘　　要：提出1种混合使用模拟退火和竞赛规则选择算子的改进GEP算法-GEPMS算法。以E-Dragon软件计算、经RM算法筛选得到的7个RDF描述符作为自变量,以抗HIV-1活性IC_(50)值作为因变量,基于GEPMS算法建立关于48种喹诺酮羧酸类化合物的HIV-1整合酶抑制剂活性的QSAR模型。与GEP、GEPSA和v-SVM算法建立的QSAR模型进行比较,本文模型、GEP、GEPSA和v-SVM模型对训练集的计算结果,决定系数R^2分别为0.9667、0.9624、0.9348和0.9711,对验证集的预测结果R^2则分别为0.9565、0.8974、0.9124和0.7656,表明本文的GEPMS模型具有最佳的泛化能力,算法的改进效果明显。In the paper, the improved GEP algorithm named GEPMS（GEP with multi-selection） which uses tournament selection and simulated annealing algorithm, has been proposed. The QSAR models have been developed, employing radial distribution function （RDF） descriptors calculated by E-Dragon software with a set of 48 compounds for quinolone carbonylic acids as HIV-1 intcgrasc inhibitors. And seven RDF descriptors have been got by using the replacement method （RM） as feature selection （descriptor selection）. GEP, GEPSA, GEPMS, and SVM models have been established For the training set, square correlation coefficient（R2） are 0.9667, 0.9624, 0.9348 and 0.9711 respectively. For the test, R2 are 0.9565, 0.8974, 0.9124 and 0.7656 respectively. The results show that GEPMS mode! has a good prediction ability. The effect of improvement for GEPMS is significant.

关 键 词：改进基因表达式编程 整合酶抑制剂 定量结构活性关系 

分 类 号：O641.12[理学—物理化学]