Real-space refinement in PHENIX for cryo-EM and crystallography.

Area of Science:

• Structural biology
• Computational biology
• Biophysics

Background:

• Accurate atomic models are crucial for understanding protein function.
• Low-resolution data presents challenges in model building and refinement.
• Existing refinement methods may not fully leverage all available structural information.

Purpose of the Study:

• To implement and evaluate a fast real-space refinement protocol within the PHENIX suite.
• To enable efficient identification of optimal data-restraint weights.
• To improve the quality of atomic models, particularly those derived from low-resolution cryo-electron microscopy (cryo-EM) data.

Main Methods:

• Utilized a simplified refinement target function in phenix.real_space_refine for rapid calculations.
• Incorporated secondary-structure, rotamer-specific, and symmetry restraints alongside covalent geometry.
• Re-refined 385 Protein Data Bank models obtained from cryo-EM at resolutions of 6 Å or better.

Main Results:

• Achieved very fast calculation times due to the simplified target function.
• Successfully identified optimal data-restraint weights during routine refinement with minimal runtime cost.
• Demonstrated significant improvements in the quality of atomic models and their fit to target maps.

Conclusions:

• The phenix.real_space_refine program offers an efficient method for improving atomic models, especially with low-resolution data.
• Integrating diverse restraints enhances model accuracy and reliability.
• This approach is valuable for refining cryo-EM structures and advancing structural biology.