Assessment of CASP7 predictions for template-based modeling targets

Citations Over TimeTop 1% of 2007 papers

Abstract

This manuscript presents the assessment of the template-based modeling category of the seventh Critical Assessment of Techniques for Protein Structure Prediction (CASP7). The accuracy of predicted protein models for 108 target domains was assessed based on a detailed comparison between the experimental and predicted structures. The assessment was performed using numerical measures for backbone and structural alignment accuracy, and by scoring correctly modeled hydrogen bond interactions in the predictions. Based on these criteria, our statistical analysis identified a number of groups whose predictions were on average significantly more accurate. Furthermore, the predictions for six target proteins were evaluated for the accuracy of their modeled cofactor binding sites. We also assessed the ability of predictors to improve over the best available single template structure, which showed that the best groups produced models closer to the target structure than the best single template for a significant number of targets. In addition, we assessed the accuracy of the error estimates (local confidence values) assigned to predictions on a per residue basis. Finally, we discuss some general conclusions about the state of the art of template-based modeling methods and their usefulness for practical applications.

Related Papers

• → Physician-Friendly Machine Learning: A Case Study with Cardiovascular Disease Risk Prediction(2019)71 cited
• → Artificial Intelligence, Machine Learning, and Medicine: A Little Background Goes a Long Way Toward Understanding(2021)29 cited
• → Application of Machine Learning in Animal Disease Analysis and Prediction(2020)26 cited
• → Sentiment Analysis by Using Supervised Machine Learning and Deep Learning Approaches(2020)3 cited
• → Breakdown of Machine Learning Algorithms(2022)1 cited