Related Experiment Videos
Fast detection of common geometric substructure in proteins
L P Chew1, D Huttenlocher, K Kedem
1Department of Computer Science, Cornell University, Ithaca, New York 14853, USA.
Summary
We developed a new method to find common 3D protein substructures by comparing alpha-carbon backbones. This approach accurately identifies similar protein shapes and domains, overcoming limitations of existing methods.
Area of Science:
- Structural bioinformatics
- Computational biology
- Protein structure analysis
Background:
- Identifying common three-dimensional (3D) substructures in proteins is crucial for understanding protein function and evolution.
- Existing methods for comparing protein shapes often struggle with noise, outliers, and identifying non-contiguous domains, or face computational challenges.
Purpose of the Study:
- To develop a novel, robust, and computationally efficient method for identifying common 3D substructures between proteins.
- To propose a geometric representation and similarity measure that excels at detecting both contiguous and non-contiguous protein domains.
Main Methods:
- Representing protein backbone chains as sequences of unit vectors derived from alpha-carbon pairs.
- Defining a similarity measure based on the root mean squared (RMS) distance between corresponding orientation vectors.
- Applying rigid 3D motions to align and compare substructures.
Main Results:
- The proposed orientation vector-based similarity measure is robust against noise and outliers.
- The method effectively identifies common contiguous substructures and the more challenging problem of common protein domains.
- This approach avoids the computational complexity of distance matrix and contact map methods.
Conclusions:
- The new geometric representation and RMS distance measure offer a superior approach for identifying common 3D protein substructures.
- This method provides a more effective and computationally feasible alternative to existing protein shape comparison techniques.