As a coarse level description of protein structure, TOPS (Topology of protein structure) diagrams are simplified descriptions of structural topology of proteins and comprise graphical representations of SSEs. In this paper, we propose a new method to analyze the similarities/dissimilarities of protein structure of TOPS representations based on the graphical representation proposed by C. X. Yuan et al. (2003). Instead of calculating the leading eigenvalues of the matrix for graphical representation, we compute torsion of curves as the descriptor to numerically characterize TOPS strings. Using the new method, TOPS strings of the 34 protein chains in the Chew-Kedem data set are analyzed. The similarities/ dissimilarities and phylogenetic tree of these protein structures verify the validity of our method.

Key words: TOPS; Torsion; Difference; Similarity.