论文部分内容阅读
基于VHSE(Principal component score vector of hydrophobic,steric,and electronic properties)结构表征方法,采用支持向量机结合遗传算法变量筛选技术,分别建立B*5701和B*5801多肽亲和活性的分类预测模型,其最优线性模型内部验证的灵敏度(Sensitivity,Sen)、特异性(Specificity,Spe)、接受者操作特征曲线下面积(Area under receiver operating characteristics curve,AUC)和马休斯相关系数(Matthews coefficient of correlation,MCC)分别为77.29%、93.99%、93.02%、67.65%(B*5701)和78.08%、89.62%、88.34%、64.73%(B*5801);外部验证的Sen、Spe、AUC和MCC分别为80.02%、94.53%、94.62%、72.09%(B*5701)和77.43%、90.79%、87.98%、66.20%(B*5801)。依据最优模型,分别对B*5701和B*5801配体的亲和特性进行了细致的比较和分析,研究结果可为Abacavir的HLA-B*5701限制性药物毒副作用(Serious Adverse Drug Reactions,SADR)机理研究提供重要的参考依据。
Based on the structural characterization of VHSE, a classification prediction model of affinity activity of B * 5701 and B * 5801 polypeptides was established by using support vector machine combined with genetic algorithm and variable screening. Sensitivity (Sen), Specificity (Spe), Area under receiver operating characteristic curve (AUC) and Matthews coefficient of correlation , MCC) were 77.29%, 93.99%, 93.02%, 67.65% (B * 5701) and 78.08%, 89.62%, 88.34%, 64.73% Were 80.02%, 94.53%, 94.62%, 72.09% (B * 5701) and 77.43%, 90.79%, 87.98%, 66.20% (B * 5801) respectively. According to the optimal model, the affinity characteristics of the B * 5701 and B * 5801 ligands were compared and analyzed respectively. The results of this study can be described as the HLA-B * 5701 restricted drug side effects of Abacavir (Serious Adverse Drug Reactions, SADR) mechanism to provide an important reference.