摘要
在三维投影的基础上 ,对投影的边缘进行了描述 ,并以其进行了定量结构活性相关研究 .实验结果表明 ,与 Am指数、指示变量和量化参数相结合 ,可使其数学模型得到明显提高 .通过人工神经网络对硝基苯类化合物的毒性进行预测 ,结果令人满意 .
In this paper, the periphery of projection of a molecule is described based on three-dimensional projection, and the new invariable was used for quantitative structure-activity relationship studies. Much better results were obtained, if Am indexes, indicated index and quantum-chemical parameters were combined with the new index. If on the new parameter, periphery of projection of a molecule, the correlation coefficient is 0.939; whereas the correlation coefficient is 0.954 because the addition of the new index. Artificial neural networks were performed to predict the toxicity of nitrobenzenes, and the satisfactory result, the correlation coefficient of 0.986, were achieved.
出处
《高等学校化学学报》
SCIE
EI
CAS
CSCD
北大核心
2004年第4期622-626,共5页
Chemical Journal of Chinese Universities
基金
国家自然科学基金 (批准号 :2 0 0 770 2 6)资助