Formalizing atom-typing and the dissemination of force fields with foyer
Published in Computational Materials Science, 2019
Abstract: A key component to enhancing reproducibility in the molecular simulation community is reducing ambiguity in the parameterization of molecular models used to perform a study. Ambiguity in molecular models often stems from inadequate usage documentation of molecular force fields and the fact that force fields are not typically disseminated in a format that is directly usable by software. Specifically, the lack of a generally applicable scheme for the annotation of the rules of a particular force field and a general purpose tool for performing automated parameterization (i.e., atom-typing) based on these rules, may lead to errors in model parameterization that are not easily identified. Here, we present Foyer, an open-source Python tool that enables users to define and apply force field atom-typing rules in a format that is both human- and machine-readable and provides a framework for force field dissemination, thus eliminating ambiguity in atom-typing and improving reproducibility. Foyer defines force fields in an XML format, where SMARTS strings are used to define the chemical context of a particular atom type and “overrides” are used to set rule precedence, rather than a rigid hierarchical scheme. Herein we describe the underlying methodology and force field annotation scheme of the Foyer software, demonstrate its application in several use-cases, and discuss specific aspects of the Foyer approach that are designed to improve reproducibility.