| Summary: | 碩士 === 國立東華大學 === 資訊工程學系 === 98 === It is well-known that proteins are the functional products of genes. The function of a protein is mainly determined or inferred from its three-dimensional structure, which is much better than corresponding amino acid residue sequence. Furthermore, the three-dimensional structure of a protein can provide sufficient information for the dis-crimination between remote homology and analogy of proteins, which is undetectable by using sequence alignment methods. In this thesis, we present a protein structure comparison method towards aligning a pair of three-dimensional protein structures by using graph theory to quantitatively measure the global similarity between two protein structures. We first transfer a protein into a labeled graph concentrating at residue level and capture sufficient information of several specific structural prop-erties and physicochemical characteristics of amino acid residues to represent the three-dimensional structure of a protein. For two graphs that correspond to two pro-tein structures, we then find their maximum common induced subgraph for measuring the structural similarity between the two proteins. Experimental results show that our method is comparable to the representatives of the major types of comparison of three-dimensional protein structures. In other words, this graph-based approach provides a new alternative for measuring protein structural similarity.
|
|---|
