Related Experiment Video
Updated: May 6, 2026
16:41
A Protocol for Computer-Based Protein Structure and Function Prediction
Published on: November 3, 2011
70.1K
Fast protein fragment similarity scoring using a Binet-Cauchy kernel.
Frédéric Guyon1, Pierre Tufféry
1Univ Paris Diderot, Sorbonne Paris Cité, Molécules Thérapeutiques in Silico, UMR 973, F-75205 Paris, France, INSERM, U973, F-75205 Paris, France and Univ Paris Diderot, Ressources Parisiennes de Bioinformatique Structurale, F-75205 Paris, France.
Bioinformatics (Oxford, England)
|October 30, 2013
Summary
A new Binet-Cauchy kernel score offers a faster, more accurate method for assessing protein structure similarity. This score improves upon root mean square deviation (RMSD) by providing length-independent statistics and better discrimination for protein fragments.
Area of Science:
- Computational Biology
- Structural Bioinformatics
- Biophysics
Background:
- Assessing protein structure similarity is crucial for understanding protein evolution and function.
- Current methods struggle with local structure analysis and statistical significance.
- The increasing volume of protein structural data necessitates efficient similarity measures.
Purpose of the Study:
- To introduce a novel scoring method for protein structure similarity.
- To address limitations of existing methods, particularly the root mean square deviation (RMSD).
- To enable efficient large-scale mining of protein structural databases.
Main Methods:
- Development of a new score based on the Binet-Cauchy kernel.
- Normalization of the score to range from -1 (mirror conformations) to 1 (identical conformations).
- Statistical analysis demonstrating length-independent properties and improved discrimination.
Main Results:
- The Binet-Cauchy kernel score provides normalized similarity values between -1 and 1.
- The score achieves length-independent statistics, even for short protein fragments.
- It demonstrates superior performance in discriminating medium-range root mean square deviation (RMSD) values compared to traditional RMSD.
Conclusions:
- The new score is simpler, faster to compute, and more effective than RMSD for local structure analysis.
- It facilitates the search for both similar and mirror image protein conformations.
- The score enables large-scale mining of protein structures and aids in deciphering protein evolution.
