In nearly every cellular process, macromolecular assemblies play critical roles, and as such, understanding the structure and function of these complexes is critical. Thanks to a number of technical advances, electron cryomicroscopy (cryoEM) has seen a tremendous growth in imaging macromolecular assemblies at near atomic resolutions. However, map interpretation and model building directly from density without the use of structural templates is still a challenge. In previous work, we developed Pathwalking, a semi-automatic toolkit that calculates potential protein structures directly from cryoEM density map. Pathwaking, based on the Traveling Salesman Problem (TSP), calculates paths (i.e. protein fold) through the density without using any sequence or structure constraints. Rather, a TSP solver is used to find a path through a set of pseudoatoms by optimizing distances between the pseudoatoms such that they are representative of Cα-Cα distances. In the initial testing of Pathwalking, reasonable first-approach models, i.e. models with the correct overall fold, were derived directly from the density map with limited user intervention. However, Pathwalking was not designed to directly consider protein chemistry or density map constraint, and as such, non- protein like connections were sometimes observed and required the user to correct. In our latest version of Pathwalking, we have improved and automated the model building protocol, including identifying secondary structures assignment, pseudoatom placement, path evaluation and refinement. Additionally, Pathwalking now allows for modeling multiple chains in a density map simultaneously. Furthermore, Pathwalking features an automated sequence registration step, as well as a C-alpha to all-atom model routine to complete the final model. With model building and refinement nearly completely automated, users can now generate high-quality all-atom models directly from a cryoEM density map within a few minutes; typically Pathwalking requires less than 30 minutes for a complete run on desktop computer. In summary, Pathwalking offers a rapid and robust method for the generation of protein structure models from near-atomic resolution cryoEM density maps

Automated de novo model building in cryoEM with Pathwalking

2019 Spring National Meeting

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