【摘要】 以量化的分子结构参数和实验结果为依据,运用模式识别技术、多元线性回归和人工神经网络研究了目标分子与DNA相互作用的主要影响因素,建立了准确性较高的2个键合常数预测模型和1个作用模式预测模型.初次量化的分子结构参数有21种,经过筛选发现其中的10种参数对相互作用有显著影响.研究结果可为抗癌药物的分子设计和筛选提供有价值的信息.更多还原

【Abstract】 On the basis of the quantified data of molecular structures and the experimental results,the pattern recognition techniques,multiple linear regression and artificial neural network were applied to studying the main influencing factors for the interaction between DNA and the target molecules,and two predictable models of binding constant and one binding mode which have a high accurate were established.First,21 structural descriptors(parameters) were used to describe a target molecule.After filtering,10 descriptors were found to affect remarkably the interaction between the target molecules and DNA.These studies could afford valuable information to design and filter molecules of anticancer drugs.更多还原