A comparative analysis of public ligand databases based on molecular descriptors

Abstract

A wide range of public ligand databases provides currently dozens of millions ligands to users. Consequently, exaustive in silico virtual screening testing with such a high volume of data is particularly expensive. Because of this, there is a demand for the development of new solutions that can reduce the number of testing ligands on their target receptors. Nevertheless, there is no method to reduce effectively that high number in a manageable amount, thus becoming this issue a major challenge of rational drug design. This article presents a comparative analysis among the main public ligand databases by measuring the quality and variations in the values of the molecular descriptors available in each one. It aims to help the development of new methods based on criteria that reduce the set of promising ligands to be tested. © 2012 Springer-Verlag.