An analysis of the redundancy of graph invariants used in chemoinformatics

摘要: Molecular descriptors play a decisive role for evaluating large virtual libraries and to predict biological or physicochemical properties of compounds. Topological indices are an important class molecular descriptors, based on the graph molecule. A major problem is that many topological considerably correlated, impeding data analysis interpretation. Also, size-dependent variance adversely affects processing by neural nets. Using random graphs as model molecules, we examine correlations abstract index with independent vertex properties. We consider making no assumptions distribution fixed number vertices in which edges selected independently. show may be strongly correlated even On other hand, uncorrelated constant Θ(1) can easily obtained within respective models.