RED: A Set of Molecular Descriptors Based on Rényi Entropy

New molecular descriptors, RED (Rényi entropy descriptors), based on the generalized entropies introduced by Rényi are presented. Topological descriptors based on molecular features have proven to be useful for describing molecular profiles. Rényi entropy is used as a variability measure to contract a feature-pair distribution composing the descriptor vector. The performance of RED descriptors was tested for the analysis of different sets of molecular distances, virtual screening, and pharmacological profiling. A free parameter of the Rényi entropy has been optimized for all the considered applications.