Detecting protein candidate fragments using a structural alphabet profile comparison approach

Area of Science:

• Computational Biology
• Structural Bioinformatics
• Protein Structure Prediction

Background:

• Protein structure modeling relies on accurate fragment prediction from sequences.
• Current methods for identifying candidate fragments often use sequence similarity or threading.
• Existing approaches face limitations in accuracy and efficiency.

Purpose of the Study:

• To develop a new profile comparison approach for identifying accurate protein fragments.
• To introduce a novel protocol for detecting position-specific fragments of varying sizes (6-27 amino acids).
• To improve the accuracy and efficiency of protein structure modeling.

Main Methods:

• Utilized predicted structural alphabet profiles, encoding local 3D shapes, for profile comparison.
• Developed a new protocol for detecting candidate fragments specific to each sequence position.
• Evaluated the approach using datasets from Critical Assessment of Techniques for Protein Structure Prediction (CASP) rounds 9 and 10.

Main Results:

• Structural alphabet profile-profile comparison effectively retrieves accurate structural fragments.
• The new protocol outperforms state-of-the-art methods in fragment accuracy and true positive rates.
• Achieved high coverage scores, demonstrating the method's comprehensive identification capabilities.

Conclusions:

• The proposed structural alphabet profile-profile comparison is a superior method for identifying protein fragments.
• This approach significantly enhances protein structure modeling by providing more accurate fragments.
• A freely available web server (http://bioserv.rpbs.univ-paris-diderot.fr/SAFrag) supports this novel methodology.