Validating Novel QSAR Descriptors for Use in Diversity Analysis

A descriptor is a mathematical function which maps chemical structures into the set of real numbers or into a real-valued vector. When functions take on single values which characterize a molecule as a whole, they can be classed as one dimensional descriptors, as can substituent parameters used to partition physical chemical effects among the substructures of which the molecules being studied are comprised. Partition coefficients (e.g., logP and ClogP) and molar refractivity (MR) are examples of ID descriptors. 2D descriptors such as atom pair and substructural fingerprints and connectivity indices explicitly incorporate contributions from molecular connectivity. 3D descriptors, in contrast, include contributions from effects (e.g., through-space interactions) which are dependent on the conformation or the position of a molecule, or both. The most commonly employed 3D descriptors are molecular fields (CoMFA).1