Protein-Ligand Docking with Protein-based and Ligand-based Structure Activity Relationships

Abstract

Protein-small molecule docking programs predict the interaction interface and energy between a given protein target and a small molecule ligand. The accuracy of docking predictions generally improve with the guidance of experimentally derived restraints. One available source of such restraints is structure-activity relationships (SARs). SARs provide information on changes in binding affinity or biological response corresponding to a chemical change in the protein and/or ligand. These chemical changes frequently refer to amino acid mutations on the protein side and functional group modifications on the ligand side. Theoretically, predicted interaction energies should correlate with SARs though in practice, this is challenging due to the difficulties in scoring protein-ligand interactions. We have previously developed RosettaLigandEnsemble (RLE), a protein-ligand docking method that simultaneously docks a congeneric ligand series to a single protein target. RLE is capable of identifying native-like binding modes for a ligand series that match the available ligand SARs. This work in progress reports on the extension of RLE to factor in SARs derived from protein mutagenesis data. The new method, ProteinLigEnsemble (PLE), is also part of the Rosetta Biomolecular Modeling Suite available at https://www.rosettacommons.org/. We have also developed protein ensemble docking features that allow for docking or screening against multiple receptor variants at the same time. We have included a proof of concept study and a tutorial for interested users.