Robust Comparison of Three-Dimensional Shapes with an Application to Protein Molecule Configurations.
PRINCETON UNIV NJ DEPT OF STATISTICS
Pagination or Media Count:
Robust statistical methods have recently been shown to have desirable properties when used for the identification of similarities and differences in shape. We present a generalization of the two-dimensional repeated median algorithm to three and higher dimensions. The extension is achieved using a duality between orthogonal and skew-symmetric matrices, which permits the definition of a median of a collection of orthogonal matrices. The methods are illustrated by comparing the predicted three-dimensional configuration of a protein molecule to a refined structure that had been found using nuclear magnetic resonance techniques. Author
- Statistics and Probability