Development of Automated Methods for Modeling Ligands in Cryo-Electron Microscopy Data

Abstract

Advances in cryo-electron microscopy (cryoEM) and deep-learning guided protein structure prediction have expedited structural studies of protein complexes. However, methods for accurately modeling ligand conformations are lacking. For my doctoral thesis, I developed computational methods to automatically determine both ligand conformation and identity in medium- to low- resolution cryoEM maps. These methods utilize both a small molecule force field in Rosetta and information in cryoEM data. First, a ligand fitting protocol EMERALD accurately predicts ligand conformations along with surrounding side chains in maps as low as 6 Å local resolution. Then, I further expand the capabilities of EMERALD to produce small molecule models similar to deposited models with 20 or more torsion angles. Finally, libraries of common ligands and lipids are screened through EMERALD to assign identities to ligand blobs of density. Together, these automatic tools can fit into cryoEM modeling pipelines for determination of protein-ligand complexes.