Dimorphite-DL: an open-source program for enumerating the ionization states of drug-like small molecules

Abstract

Small-molecule protonation can promote or discourage protein binding by altering hydrogen-bond, electrostatic, and van-der-Waals interactions. To improve virtual-screen pose and affinity predictions, researchers must account for all major small-molecule ionization states. But existing programs for calculating these states have notable limitations such as high cost, restrictive licenses, slow execution times, and poor modularity. Here, we present dimorphite-DL 1.0, a fast, accurate, accessible, and modular open-source program for enumerating small-molecule ionization states. Dimorphite-DL uses a straightforward empirical algorithm that leverages substructure searching and draws on a database of experimentally characterized ionizable molecules. We have tested dimorphite-DL using several versions of Python and RDKit on all major operating systems. We release it under the terms of the Apache License, Version 2.0. A copy is available free of charge from http://durrantlab.com/dimorphite-dl/ .

Keywords: Drug discovery; Ionization; Modeling; Protonation; Virtual screening; pH.

Fig. 1

A schematic representation of the dimorphite-DL approach. Each ionizable moiety is associated with a pK_a range (range_PKA) defined by three parameters: µ, σ, and n. The user specifies a pH range (range_pH) and pK_a precision factor (n; default: 1.0). The mean (µ) and standard deviation (σ) associated with each moiety are derived from the database of small molecules with experimentally characterized pK_a values. If range_PKA is entirely less than range_pH, dimorphite-DL outputs a deprotonated molecule. If range_PKA is entirely greater than range_pH, dimorphite-DL outputs a protonated molecule. If range_PKA and range_pH overlap, dimorphite-DL outputs both deprotonated and protonated molecules