Predicting the performance of automated crystallographic model-building pipelines

Area of Science:

• Structural biology
• Computational biology
• Biochemistry

Background:

• Protein structure determination is crucial for biological function.
• Computational pipelines are used to model protein structures from crystallographic data.
• Pipeline performance varies, making optimal selection challenging for researchers.

Purpose of the Study:

• To develop a software tool that predicts the quality of protein models generated by different software pipelines.
• To guide researchers in selecting the most efficient pipelines for protein structure modeling.
• To improve the accuracy and completeness of final protein models.

Main Methods:

• The tool predicts crystallographic quality-of-fit indicators and structure completeness.
• Predictions are based on input crystallographic data.
• The software was tested on over 2500 experimental phasing and molecular replacement data sets.

Main Results:

• The tool accurately predicts quality measures for protein structures.
• Predictions were validated across a wide range of resolutions for both experimental phasing (1.2–4.0 Å) and molecular replacement (1.0–3.5 Å) data.
• The software demonstrates high accuracy in predicting key quality metrics.

Conclusions:

• The developed tool effectively predicts protein model quality from crystallographic data.
• It enables informed selection of computational pipelines for efficient protein structure modeling.
• This facilitates the generation of high-quality, depositable protein models.