Protein structure validation by generalized linear model root-mean-square deviation prediction.

Area of Science:

• Structural biology
• Computational biology
• Bioinformatics

Background:

• Accurate protein structure determination is crucial for understanding biological function.
• Existing methods for assessing protein structure accuracy have limitations.
• Large-scale projects like CASP and CASD-NMR highlight the need for robust validation criteria.

Purpose of the Study:

• To develop a universal method for assessing protein structure accuracy.
• To combine diverse protein structure quality scores into a single, meaningful metric.
• To predict the root-mean-square deviation (RMSD) to the true structure.

Main Methods:

• Development of a generalized linear model (GLM) to integrate multiple quality scores.
• Application of the GLM-RMSD method to structural models from CASD-NMR and CASP.
• Comparison of GLM-RMSD predictions with actual RMSD values to experimentally determined structures from the Protein Data Bank (PDB).

Main Results:

• The GLM-RMSD method demonstrated high correlation coefficients (0.69 for CASD-NMR, 0.76 for CASP) between predicted and actual heavy-atom RMSDs.
• GLM-RMSD significantly outperformed individual coordinate-based quality scores in predicting accuracy.
• The method provides an intuitive measure of structural accuracy, predicting deviation from the true structure.

Conclusions:

• The GLM-RMSD method offers a more reliable approach to protein structure quality assessment.
• This approach effectively combines diverse structural quality metrics for improved accuracy prediction.
• GLM-RMSD is a valuable tool for critically assessing protein structure determination and prediction methods.