论文部分内容阅读
分子MOLMAP指数是通过递交一个分子中所有化学键描述符到Kohonen自组织映射,基于一个特定的算法所生成,其中Kohonen自组织映射已经预先由多种类型化学键所衍生的描述符进行了训练。化学键描述符由化学键的物理化学性质或(和)拓扑性质组成。MOLMAP指数是一个基于化学键类型的分子指数。若2个描述符相似的化学键激活同一个Kohonen自组织映射神经元,则2个化学键为同一类型。进一步,采用反应物与生成物的MOLMAP指数差异性描述化学反应,衍生了化学反应的MOLMAP指数。此指数能够在没有参与化学反应的化学键信息的情况下进行化学反应的相关研究,因而能够广泛地应用。本文从3个方面概述了MOLMAP指数的应用:1)在化学反应分类中的应用,例如:预测生物代谢反应分类;2)在化学反应性预测中的应用,化学反应性是指与化学反应有关的性质,例如:预测物质的变异性和抗氧化性;3)在三维数据处理中的应用,是基于MOLMAP指数算法进行的扩展,例如:预测氯酚类化合物的毒性。
The molecular MOLMAP index is generated by submitting a map of all chemical bond descriptors in a molecule to the Kohonen self-organizing map, generated on the basis of a specific algorithm, where the Kohonen self-organizing map has been previously trained on descriptors derived from various types of chemical bonds. Chemical bond descriptors consist of the physicochemical or (and) topological properties of a chemical bond. The MOLMAP index is a molecular index based on the type of chemical bond. If two chemical bonds with similar descriptors activate the same Kohonen self-organized map neurons, the two chemical bonds are the same type. Further, the MOLMAP index was used to describe the chemical reaction by difference between the reactants and the resultant, and the MOLMAP index of the chemical reaction was derived. This index enables chemical reactions to be studied without the chemical bond involved in chemical reactions, and thus can be widely used. This article summarizes the application of the MOLMAP index in three aspects: 1) application in chemical reaction classification, for example: prediction of the classification of biological metabolic reactions; 2) application in chemical reactivity prediction, chemical reactivity refers to the chemical reaction Such as predicting material variability and antioxidant activity; and 3) application in three-dimensional data processing based on the MOLMAP exponential algorithm, such as predicting the toxicity of chlorophenols.