论文部分内容阅读
利用计算机将致癌化合物样本按KLN方法作结构结点编码、分解,获得机器可识别的化合物结构片段。根据模糊数学的隶属度概念和聚类方法,计算它对化合物诱变活性作用的隶属度,以衡量片段结构对化合物诱变活性的影响,在此量化基础上作模糊聚类,得特征的活性片段类,对化合物进行特征片段的匹配及匹配特征片段的量化平均,获得化合物活性的数值度量。计算表明:隶属度值的分布合理,片段的隶属度也确实反映了它们的活性情况,特征片段聚类后更确切的反映化合物的活性。同时,作为计算样本,当数量增加,效果更显著;由于片段分布的复杂性(错位、多样),用模糊性的度量可反映其本质。本文研究表明:模糊性的度量对于研究化合物构效关系是较好的方法。
Using computer to make samples of carcinogenic compounds according to the KLN method for coding and decomposing the structural nodes, and obtaining machine identifiable structural fragments of the compounds. According to the concept of membership degree of fuzzy mathematics and the clustering method, its membership degree of the mutagenic activity of the compound is calculated to evaluate the influence of the fragment structure on the mutagenic activity of the compound. On the basis of the quantification, fuzzy clustering is performed to obtain the characteristic activity Fragments, match the signature of the compound and match the quantified average of the signature to obtain a numerical measure of the activity of the compound. The calculated results show that the distribution of membership values is reasonable and the membership of the fragments does reflect their activity. The clustering of the characteristic fragments more accurately reflects the activity of the compounds. At the same time, as the number of samples increases, the effect becomes more significant as the number increases; the measure of fuzziness can reflect its essence due to the complexity (dislocation, diversity) of fragment distributions. The research shows that the fuzziness measurement is a good method to study the structure-activity relationship of compounds.