Does the accounting of the local symmetry fragments in SMILES improve the predictive potential of the QSPR-model for Henry's law constants?

Abstract

When modeling many physicochemical, biochemical, and ecological processes, numerical data on Henry's law constants are much desired. In addition, these data are used in pharmaceuticals for the development of gaseous drugs, as well as in modeling drug–receptor interactions. Henry's law constant is an indicator of the affinity of compounds for the vapor phase and water. The local symmetry of simplified molecular input-line entry systems (SMILES) comprises compositions of identical symbols that can be represented as three ‘xyx’, four ‘xyyx’, or five symbols ‘xyzyx’. Taking account of these attributes of SMILES can improve the predictive potential of models for Henry's law constants. We updated our CORAL software using the optimal (flexible) descriptor. The updated descriptor improved the predictive potential when applied to the model for Henry's law constants. This new approach also permits fast definition of a set of pollutants that have a minimal impact on climate change and are safe from an environmental point of view.