Residue-wise local quality estimation for protein models from cryo-EM maps

Area of Science:

• Structural Biology
• Biophysics
• Computational Biology

Background:

• Cryo-electron microscopy (cryo-EM) is increasingly used to determine protein structures.
• Despite improving resolution, accurate amino acid assignment in cryo-EM density maps remains challenging.
• Existing methods may struggle with residue misassignments, including shifts along the main chain.

Purpose of the Study:

• To develop a computational method for identifying potential misassignments of residues in cryo-EM density maps.
• To assess the consistency between protein structure models and cryo-EM data likelihoods.
• To improve the accuracy and reliability of protein structure models derived from cryo-EM.

Main Methods:

• Developed a deep learning-based scoring method named DAQ (Data-model Accuracy Quotient).
• DAQ computes the likelihood of local density corresponding to various amino acids, atoms, and secondary structures.
• Assesses the consistency of amino acid assignments in protein structure models against these likelihoods.

Main Results:

• When applied to different versions of Protein Data Bank models from the same cryo-EM maps, newer models showed improved DAQ scores.
• DAQ successfully identified potential misassignment errors in a significant number of deposited protein structure models.
• The method demonstrates effectiveness in detecting inaccuracies in cryo-EM-derived protein structures.

Conclusions:

• The DAQ score provides a reliable metric for assessing the quality of amino acid assignments in cryo-EM structural models.
• This method can help researchers identify and correct errors in deposited protein structures, enhancing data integrity.
• DAQ contributes to improving the accuracy of structural biology research relying on cryo-EM data.