EVA: large-scale analysis of secondary structure prediction

Area of Science:

• Computational biology
• Structural bioinformatics
• Protein structure prediction

Background:

• The EVA (Evaluation of WWW-based Protein Structure Prediction) server has continuously monitored automatic protein structure prediction servers since June 2000.
• Over 20,000 secondary structure predictions have been collected, providing a substantial dataset for analysis.

Purpose of the Study:

• To assess the advancements in automatic protein secondary structure prediction.
• To evaluate the impact of recent improvements, particularly the use of evolutionary information, on prediction accuracy.

Main Methods:

• Continuous, objective evaluation of web-based automatic structure prediction servers via the EVA system.
• Analysis of over 20,000 secondary structure predictions collected since June 2000.
• Utilizing evolutionary information derived from protein families and improved database searches.

Main Results:

• Protein secondary structure prediction accuracy has significantly advanced, reaching approximately 76% correct residue predictions.
• Recent improvements, driven by enhanced evolutionary information, have boosted accuracy by over 4%.
• Current leading methods effectively predict secondary structure segments and perform well on predicting strands.

Conclusions:

• The recent increase in prediction accuracy is substantial and enhances the practical utility of these computational methods.
• The EVA server provides a valuable resource for tracking progress and understanding trends in protein structure prediction.
• Further advancements in prediction accuracy are anticipated, with the ultimate limit yet to be determined.