The analysis of the three-dimensional structure of objects and the identification of regions of similarity among structures arises in diverse applications and has been the subject of much study. In computational biology, the search for common structural "motifs" in a set of proteins is believed to help reveal functional relationships among different proteins and to lead to the discovery of evolutionary relations that are not evident from their linear sequence of amino acids.
We are experimenting on various matching techniques for mode-based object recognition, for pairwise protein comparison, and for fast retrieval of motifs from the Protein Data Bank.
We are also investigating the geometry of secondary structure elements that help in predicting the structures of proteins.