SAD phasing of XFEL data depends critically on the error model.

Area of Science:

• Crystallography
• Structural Biology
• Biophysics

Background:

• Serial crystallography (SX) is a powerful technique for determining protein structures.
• Single-wavelength anomalous diffraction (SAD) phasing remains a challenge in SX, often requiring thousands of diffraction patterns.
• Accurate estimation and propagation of errors in reflection intensities are crucial for successful SAD phasing.

Purpose of the Study:

• To present a nonlinear least-squares method for refining error estimates in SX data.
• To demonstrate the improved SAD phasing ability of this method.
• To enable de novo protein structure solution from SX data, even with weak anomalous signals.

Main Methods:

• A nonlinear least-squares method was developed to refine parametric expressions for estimated errors in reflection intensities.
• The method propagates error estimates from photon-counting statistics and other sources of experimental uncertainty.
• The approach was applied to SX data, incorporating terms proportional to experimental uncertainty, reflection intensity, and squared reflection intensity.

Main Results:

• The refined error estimates significantly improved SAD phasing ability in SX.
• The method enabled the autobuilding of a protein structure that had previously failed to be built.
• Successful SAD phasing was achieved even with a weak zinc anomalous signal, overcoming a major challenge in the field.

Conclusions:

• The presented nonlinear least-squares method effectively refines error estimates in SX data.
• Improved error estimation enhances SAD phasing, facilitating de novo structure determination.
• This approach offers a promising solution for solving challenging protein structures using SX and SAD phasing.