Automated Identi cation of Three-Dimensional Common Structural Features of Proteins

This paper describes an approach to automated identi cation of three-dimensional (3D) common structural features of proteins. The structure of a protein was represented by a set of secondary structure elements (SSEs) in the same manner used in our previous work, where only -helices and -strands were considered. The maximal common subgraph matching algorithm, based on a graph theoretical clique nding approach, was used to identify the 3D common structural features for a pair of proteins. The program called AIM (Automated Identi cation of 3D Motif in proteins) was developed and tested by the execution trials for nding the secondary structure segments related to the Rossmann-fold motif as a 3D common structural feature between alcohol dehydrogenase and lactate dehydrogenase, both of which are known to have the motif site. The result of a substructure search for a protein structure database using the 3D structural feature that was identi ed will also be discussed.