GLASS: a tool to visualize protein structure prediction data in three dimensions and evaluate their consistency

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Protein Structure Prediction

Background:

• Homology modeling is limited for proteins lacking sequence similarity to known structures.
• Novel methods like fold recognition and secondary structure prediction show promise but require robust evaluation.
• Assessing the accuracy of ab initio structure predictions is often difficult.

Purpose of the Study:

• To develop a platform for integrating and evaluating diverse protein structure prediction methods.
• To provide tools for comparing results from multiple prediction sources and experimental data.
• To facilitate the assessment of novel structure prediction and evaluation techniques.

Main Methods:

• Development of GLASS, a general platform for reading, visualizing, and comparing prediction results.
• Integration of multiple prediction methods (secondary structure, fold recognition, etc.) and experimental data.
• Three-dimensional projection of prediction results and comparison with known protein structure distributions.

Main Results:

• GLASS enables comprehensive comparison and evaluation of protein structure predictions.
• The platform facilitates consistency checks across different prediction methods.
• GLASS aids in assessing the reliability of predicted structural models by comparing them to real protein structures.

Conclusions:

• GLASS serves as a valuable 'workbench' for structural predictions, aiding both theoreticians and experimentalists.
• The platform enhances the evaluation of novel protein structure prediction methods.
• Integrating diverse data sources and prediction tools improves the reliability assessment of protein models.