De novo main-chain modeling for EM maps using MAINMAST.

Area of Science:

• Structural Biology
• Computational Biology
• Biophysics

Background:

• Cryo-electron microscopy (cryo-EM) is increasingly used to determine protein structures at near-atomic resolution.
• Building accurate, full-atom protein models from cryo-EM maps, especially those around 4-5 Å, remains challenging and time-consuming.

Purpose of the Study:

• To introduce MAINMAST, a novel, fully automated method for de novo protein structure modeling from cryo-EM maps.
• To evaluate MAINMAST's performance against existing software for global protein structure modeling and fragment building.

Main Methods:

• MAINMAST directly traces protein main-chains within cryo-EM density maps.
• It identifies alpha-carbon (Cα) positions by representing them as tree-graph structures.
• The method was tested on simulated and experimentally determined cryo-EM datasets.

Main Results:

• MAINMAST demonstrated superior performance in building global protein structure models compared to existing software on both simulated (5 Å) and experimental (2.6-4.8 Å) datasets.
• The method successfully built missing protein fragments (11-161 residues) with an average Root Mean Square Deviation (RMSD) of 2.68 Å.

Conclusions:

• MAINMAST offers a significant advancement in automating the process of protein model building from cryo-EM data.
• The method enhances the efficiency and accuracy of structural modeling, facilitating faster insights in structural biology.