Integrated Protocol of Protein Structure Modeling for Cryo-EM with Deep Learning and Structure Prediction

Area of Science:

• Structural biology
• Biophysics
• Computational biology

Background:

• Cryo-electron microscopy (cryo-EM) is crucial for determining protein structures.
• Accurate protein main-chain tracing in cryo-EM maps remains challenging, even at high resolutions.

Approach:

• Developed DeepMainmast, a deep learning method for protein main-chain tracing using local map features.
• Integrated DeepMainmast with AlphaFold2 to leverage de novo density tracing and structure prediction.
• The protocol accurately assigns chain identity for homo-multimer structure models.

Key Points:

• DeepMainmast enhances main-chain tracing in cryo-EM maps by analyzing local atomic features.
• Combining deep learning tracing with AlphaFold2 prediction yields superior structural modeling accuracy.
• The method effectively identifies chain identities in homo-multimeric protein complexes.

Conclusions:

• DeepMainmast offers a robust solution for challenging cryo-EM map interpretation.
• The synergistic approach with AlphaFold2 advances the accuracy of protein structure modeling.
• This protocol improves the analysis of homo-multimeric protein structures from cryo-EM data.