Identification of rogue datasets in serial crystallography

Area of Science:

• Structural Biology
• Crystallography
• Biophysics

Background:

• Serial crystallography enables data collection from small crystals using advanced X-ray techniques.
• Merging datasets enhances data completeness but requires crystals to be isomorphous (structurally similar).
• Identifying and excluding non-isomorphous datasets is crucial for accurate structural determination.

Purpose of the Study:

• To develop a precise method for identifying and excluding non-isomorphous datasets in serial crystallography.
• To improve the reliability of structural models derived from serial crystallography data.
• To prevent bias in refined models caused by the inclusion of dissimilar datasets.

Main Methods:

• A precise calculation of the correlation coefficient (CC1/2) is employed, avoiding random subset assignment.
• CC1/2 is averaged over resolution shells for more robust indicator values.
• The selection procedure's efficacy is validated by correlating observed and model-calculated intensities.

Main Results:

• The developed method accurately identifies non-isomorphous datasets.
• Inclusion of non-isomorphous data can bias refined structural models.
• Averaging CC1/2 over resolution shells provides a more reliable indicator than overall CC1/2.

Conclusions:

• The precise CC1/2 calculation method is essential for ensuring data quality in serial crystallography.
• Excluding non-isomorphous datasets prevents structural model bias.
• This approach enhances the accuracy and reliability of macromolecular structures determined by serial crystallography.