Intrinsic graph distances compared to euclidean distances for correspondent graph embedding
Academic Article
Overview
Additional Document Info
View All
Overview
abstract
Chemical structures of organic compounds are characterized numerically by a variety of structural descriptors computed either from the molecular graph or from the three-dimensional (3D) molecular geometry. Extensive use of such structural descriptors or topological indices has been made in drug design, screening of chemical databases, similarity and diversity assessment, and quantitative structure-activity relationships. In recent years a large variety of topological indices were derived from different sorts of graph distance functions which have been considered to characterize the molecular shape and structure. These include not only the shortest-path distance but also the resistance distance and the quasi-Euclidean distance. A comparison is made between five intrinsic graph distance functions and the geometric distance for a set of benzenoid hydrocarbons. Overall, a very good correlation is obtained for all graph distances, indicating that the graph descriptors derived from them capture some part of the 3D information of the molecular structure.