论文部分内容阅读
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。
A quantitative structure-activity relationship research method based on stochastic resonance theory is proposed. Stochastic resonance exists in the noise energy transfer to the signal, thereby enhancing the signal to noise ratio mechanism is an effective way to deal with noise signals. In this paper, the stochastic resonance algorithm is applied to the quantitative structure-activity relationship study of a group of anti-AIDS drugs. The error and redundancy variables which are ubiquitous among the descriptors of molecular structure are noise, and the correlation between the stochastic resonance algorithm and the activity of the studied compounds Good molecular structure descriptors as signals, through the optimization of nonlinear system parameters, to achieve a random resonance, to inhibit the error, and enhance the purpose of the signal. The statistical properties of the proposed four-variable model are: n = 34, R ~ 2 = 0.9601, s = 0.29, Q ~ 2 = 0.9508, F = 168.62. The results show that the stochastic resonance method is an effective way to improve the performance of quantitative structure-activity relationship model.