DAQplugin: Deep Learning based Real-time Model Evaluation Plugin for ChimeraX

Area of Science:

• Structural Biology
• Computational Biology
• Biophysics

Background:

• Cryo-electron microscopy (cryo-EM) is increasingly used for protein structure determination.
• Protein structure modeling from cryo-EM data often faces challenges like residue misassignments and sequence register shifts, especially in low-density regions.

Purpose of the Study:

• To introduce DAQplugin, a ChimeraX plugin for real-time, deep-learning-based validation of protein models against cryo-EM density maps.
• To enable early detection and correction of modeling errors during the structure building process.

Main Methods:

• Development of DAQplugin, a ChimeraX plugin utilizing a deep-learning-based residue-wise model quality (DAQ) score.
• Integration of real-time validation directly into the interactive model building and refinement workflow.
• Efficient computation designed for standard CPUs, eliminating the need for GPU hardware.

Main Results:

• DAQplugin provides real-time, residue-wise validation of protein models within an interactive modeling environment.
• The plugin identifies potential modeling errors and sequence register shifts, offering guidance for corrections.
• Validation can be performed on standard laptops during model building, fitting, and refinement.

Conclusions:

• DAQplugin is the first tool offering real-time deep-learning-based validation for protein models against cryo-EM maps in an interactive setting.
• It facilitates more accurate interpretation of cryo-EM density maps and enhances the reliability of protein structure models.
• The tool improves the efficiency and accuracy of cryo-EM structure determination workflows.