XRRpred: accurate predictor of crystal structure quality from protein sequence

Area of Science:

• Structural Biology
• Bioinformatics
• Computational Biology

Background:

• X-ray crystallography yields most protein structures, but quality varies (resolution, R-free).
• Accurate prediction of protein structure quality from sequence is crucial for applications like drug design and molecular docking.
• Existing tools predict crystallizability but not structure quality metrics.

Purpose of the Study:

• To introduce XRRpred, the first predictor of protein structure resolution and R-free values directly from amino acid sequences.
• To provide an accurate method for assessing potential protein structure quality prior to experimental determination or analysis.

Main Methods:

• XRRpred utilizes original sequence profiles, hand-crafted features, and selected regressors.
• Modern resampling techniques were employed for robust model training and validation.
• Performance was evaluated on an independent test dataset.

Main Results:

• XRRpred accurately predicts protein structure resolution and R-free values.
• Predictions correctly model the relationship between resolution and R-free.
• XRRpred outperforms alternative methods, including crystallization propensity predictors and alignment-based approaches.

Conclusions:

• XRRpred is a valuable tool for predicting protein structure quality from sequence.
• The webserver offers batch predictions and result visualization.
• Accurate quality prediction facilitates more reliable use of protein structures in various applications.