Extending the resolution and phase-quality limits in automated model building with iterative refinement.

Area of Science:

• Structural Biology
• Crystallography
• Biophysics

Background:

• Automated model building in crystallography relies on accurate phase information.
• Current methods for incorporating prior phase information in refinement have limitations.
• Single-wavelength anomalous diffraction (SAD) experiments provide valuable anomalous signal for phase determination.

Purpose of the Study:

• To evaluate a novel multivariate likelihood function for SAD data that directly utilizes prior phase information.
• To compare the performance of the new SAD function against existing refinement targets.
• To assess the impact on automated model building and iterative refinement.

Main Methods:

• Application of a multivariate likelihood function incorporating prior phase information from SAD experiments.
• Testing the SAD function on diverse SAD datasets with varying resolution and anomalous signal.
• Comparison with currently used refinement functions in automated model building workflows.

Main Results:

• The SAD multivariate likelihood function extends resolution and phase-quality limits for successful automated model building.
• This approach reduces overfitting during the refinement procedure.
• Consistent improvement in model quality and the number of residues built compared to current targets.

Conclusions:

• The multivariate likelihood function offers a more theoretically sound method for utilizing prior phase information in SAD refinement.
• This advancement significantly enhances the capabilities of automated model building in crystallography.
• The proposed method leads to higher quality crystallographic models.