Predicting p<i>K</i>_a Using a Combination of Semi-Empirical Quantum Mechanics and Radial Basis Function Methods

Abstract

The acid dissociation constant (p<i>K</i>_a) has an important influence on molecular properties crucial to compound development in synthesis, formulation, and optimization of absorption, distribution, metabolism, and excretion properties. We will present a method that combines quantum mechanical calculations, at a semi-empirical level of theory, with machine learning to accurately predict p<i>K</i>_a for a diverse range of mono- and polyprotic compounds. The resulting model has been tested on two external data sets, one specifically used to test p<i>K</i>_a prediction methods (SAMPL6) and the second covering known drugs containing basic functionalities. Both sets were predicted with excellent accuracy (root-mean-square errors of 0.7-1.0 log units), comparable to other methodologies using a much higher level of theory and computational cost.