Molecular docking is a computational process in which the non-covalent binding of a set of molecules is predicted by generating conformations of the inputs, usually a protein and ligand. Typically, docking is calculated using only the information of one protein and one ligand; however, previous work has demonstrated enhanced performance by understanding that distinct ligands will bind to a receptor in similar ways. Proprietary software limits both the customization and access to state-of-the-art science; therefore, we restructure a previously developed pipeline, ComBind, to use only open-source tools and a deep learning docking software, GNINA, to determine the docked structure of a series of ligands simultaneously. Features, like hydrogen-bonding and hydrophobic contacts, are determined for every docked pose of each ligand and the similarities of these features are determined between pairs of poses. We can then compute the likelihood of a given pair's similarity using the ratio of the distribution of all pose similarities and the distribution of native-like pose similarities. Our open-source docking framework's performance is compared to the previously developed framework, ComBind. We further investigate novel uses of this docking pipeline, such as ensemble docking.

Open-ComBind: An open-source docking pipeline harnessing pairwise pose statistics

Where thousands of chemistry professionals meet to share ideas and advance scientific and technical knowledge. Spring 2023 abstracts, posters, and presentations will launch at the end of March.

ACS Spring 2023

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.