CryoTEN: Efficiently Enhancing Cryo-EM Density Maps Using Transformers.

Area of Science:

• Structural biology
• Biophysics
• Computational biology

Background:

• Cryogenic electron microscopy (cryo-EM) is vital for determining macromolecular structures.
• Cryo-EM density maps often suffer from noise and missing data, hindering accurate protein structure building.
• Existing map sharpening techniques have limitations in efficiently improving cryo-EM map quality.

Purpose of the Study:

• To introduce CryoTEN, a novel deep learning method for enhancing cryo-EM density maps.
• To evaluate CryoTEN's effectiveness in improving map quality and facilitating de novo protein structure modeling.
• To compare CryoTEN's performance against state-of-the-art methods in terms of accuracy, speed, and computational resource usage.

Main Methods:

• CryoTEN utilizes a three-dimensional U-Net style transformer architecture.
• The model was trained on 1,295 cryo-EM maps and their corresponding simulated maps.
• Performance was validated using an independent test set of 150 cryo-EM maps.

Main Results:

• CryoTEN significantly enhances the quality of cryo-EM density maps.
• Protein structures modeled from CryoTEN-processed maps show substantially improved quality.
• CryoTEN achieves competitive map quality enhancement while being over 10 times faster and requiring less GPU memory than existing methods.

Conclusions:

• CryoTEN is an effective and efficient tool for improving cryo-EM density maps.
• The method facilitates more accurate de novo protein structure determination from cryo-EM data.
• CryoTEN represents a significant advancement in cryo-EM data processing and structural biology.