Related Experiment Videos
'Hybrid protein model' for optimally defining 3D protein structure fragments
1Equipe de Bioinformatique Génomique et Moléculaire, INSERM U436, Université Paris 7, case 7113, 2, place Jussieu, 75251 Paris cedex 05, France. debrevern@urbb.jussieu.fr
Bioinformatics (Oxford, England)
|February 14, 2003
Summary
This study refines a hybrid protein model for homology modeling by optimizing protein fragment libraries. The improved method generates a non-redundant repertory of 3D protein structure fragments for enhanced molecular modeling applications.
Area of Science:
- Computational Biology
- Structural Bioinformatics
- Protein Modeling
Background:
- The hybrid protein model utilizes a structural alphabet of 16 Protein Blocks (PBs) to learn protein local folds.
- Previous work established a method for defining contiguous 3D protein structure fragments based on this model.
Purpose of the Study:
- To develop an automated process for defining a repertory of 3D protein structure fragments for homology modeling.
- To present improvements to the existing hybrid protein model and its fragment library generation.
Main Methods:
- Introducing 'baby training': starting with large fragments and progressively reducing their size.
- Implementing the deletion of redundant parts within the hybrid protein's fragment database.
- Learning local folds by identifying and slightly modifying similar patterns within the hybrid protein.
Main Results:
- An optimized hybrid protein construction method was developed.
- A non-redundant databank of contiguous 3D protein structure fragments was created.
- The refined repertory of fragments is suitable for molecular modeling.
Conclusions:
- The improved method enhances the creation of a useful repertory of 3D protein structure fragments.
- This optimized fragment library is a valuable tool for homology modeling and other molecular modeling applications.