应用随机共振研究一组抗癌药物定量结构活性相关

Quantitative structure-activity relationship study on a group of HEPT derives by applying stochastic resonance

提出了一种基于随机共振理论的定量构效关系研究方法。随机共振现象中存在着将噪声的能量转移给信号,从而提高信号信噪比的机制,是一种处理噪声信号的有效方法。本文将随机共振算法应用于一组抗艾滋病药物的定量构效关系研究,以普遍存在于分子结构描述子之中的误差和冗余变量为噪声,而将与所研究化合物活性之间相关性较好的分子结构描述子作为信号,通过优化非线性系统的参数,实现了随机共振,达到了抑制误差、增强信号的目的。所建四变量模型的统计特性为：n=34,R^2=0.9601,s=0.29,Q^2=0.9508,F=168.62。结果表明,随机共振方法是一种提高定量构效关系模型性能的有效途径。

Based on the theory of stochastic resonance （ SR）, a new approach to quantitative structure-activity relationship （QSAR） studies is proposed. SR, a well-known phenomenon wherein the energy of noise of a noisy signal can be transferred to the concerned signal in a nonlinear system, has been proved to be an efficient approach to the problem of noise in several scientific fields. In this paper, SR is applied to deal with errors existing in descriptors for a group of HEPT anti-HIV drugs. The errors and the redundant descriptors were regarded as noise, but the descriptors which are relevant to the studied activity as signals in the new algorithm. The parameters of nonlinear system associated with SR were optimized according to the correlation coefficient between descriptors and the studied bioaetivity. The model obtained by SR for the anti-HIV activity of theses drugs, which is composed of 4 variables, gives the following statistics ：n = 34 ,R^2 = 0. 9601 ,s = 0. 29, Q^2 = 0. 9508, F = 168.62. It is demonstrated that SR is a potential tool to improve the predictive ability of QSAR models.