论文部分内容阅读
本文对镇痛、抗胆碱能、抗抑郁、抗组织胺、安定和抗帕金森氏症等六种药理类别的80个药物设计了一个“编码超几何结构”(coding hypergeometricstructure),然后对这80个药物逐一进行结构迭加,用拓扑指数计算每个药物在编码超几何结构的标码位置上的价分子连接值,从而获得药物分子特征的数据矩阵,再用主成分分析法对上述高维矩阵进行非迭代型映照,根据解释方差的预定要求,将其压缩到四维空间中,并计算每个药物的因子坐标。最后采用 Bayes 逐步判别法对80个药物进行六种药理活性的判别分类,判对69个,判对率为86.25%。
This article devised a coding hypergeometrics structure for 80 drugs in six pharmacological categories, analgesic, anticholinergic, antidepressant, antihistamine, diazepam and antiparkinsonism, Eighty drugs were superimposed one by one, and the topological index was used to calculate the valence molecular linkage value of each drug in the superscript code position of the superstructure to obtain the data matrix of the molecular characteristics of the drug. The principal component analysis Non-iterative mapping of the matrix of dimensions is performed, which is compressed into four-dimensional space according to the predetermined requirement for interpretation of variance and the factor coordinates of each drug are calculated. Finally, Bayes discriminant method was used to classify and classify sixty pharmacological activities of 80 drugs, with 69 judgments of 86.25%.