Fast and automated protein-DNA/RNA macromolecular complex modeling from cryo-EM maps

Area of Science:

• Structural biology
• Computational biology

Background:

• Cryo-electron microscopy (cryo-EM) enables 3D reconstruction of macromolecular structures, crucial for understanding biological mechanisms and diseases.
• Accurate model building of large macromolecular complexes from cryo-EM data is challenging and time-consuming.

Purpose of the Study:

• To develop an advanced computational pipeline for automated model building from cryo-EM maps.
• To improve the accuracy and efficiency of macromolecular structure determination using artificial intelligence.

Main Methods:

• Development of DeepTracer-2.0, an AI-based pipeline for building amino acid and nucleic acid backbones from cryo-EM maps.
• Prediction of best-fitting residues based on side-chain density in the cryo-EM map.
• Benchmarking on independent experimental cryo-EM maps of protein-DNA/RNA complexes.

Main Results:

• DeepTracer-2.0 demonstrated improved accuracy and efficiency in building macromolecular models.
• The AI pipeline successfully reconstructed backbones and predicted residues from cryo-EM density maps.
• Validation on independent datasets confirmed the method's robustness.

Conclusions:

• DeepTracer-2.0 offers a significant advancement in automated macromolecular modeling from cryo-EM data.
• The pipeline has the potential to accelerate research in molecular biomedicine and drug discovery.
• The tool is accessible via a web portal, facilitating wider adoption by researchers.