Multiconformation, Density Functional Theory-Based pKa Prediction in Application to Large, Flexible Organic Molecules with Diverse Functional Groups

We consider the conformational flexibility of molecules and its implications for micro- and macro-pKa. The corresponding formulas are derived and discussed against the background of a comprehensive scientific and algorithmic description of the latest version of our computer program Jaguar pKa, a density functional theory-based pKa predictor, which is now capable of acting on multiple conformations explicitly. Jaguar pKa is essentially a complex computational workflow incorporating research and technologies from the fields of cheminformatics, molecular mechanics, quantum mechanics, and implicit solvation models. The workflow also makes use of automatically applied empirical corrections which account for the systematic errors resulting from the neglect of explicit solvent interactions in the algorithm’s implicit solvent model. Applications of our program to large, flexible organic molecules representing several classes of functional groups are shown, with a particular emphasis in illustrations laid on drug-like molecules. It is demonstrated that a combination of aggressive conformational search and an explicit consideration of multiple conformations nearly eliminates the dependence of results on the initially chosen conformation. In certain cases this leads to unprecedented accuracy, which is sufficient for distinguishing stereoisomers that have slightly different pKa values. An application of Jaguar pKa to proton sponges, the pKa of which are strongly influenced by steric effects, showcases the advantages that pKa predictors based on quantum mechanical calculations have over similar empirical programs.