Epik Background

Epik predicts the populations of an ionizable ligand’s protonation states at a target pH. Like previous versions of Epik, state populations are calculated from the microscopic pK_a values of each state. However, unlike previous versions, which used the Hammett-Taft approach to estimate pK_a values based on highly parameterized linear free energy relationships, Epik uses atomic graph convolutional neural networks (GCNN) to predict pK_a values.

Epik Evaluation