Study Of Protein Structure Classification

Bioinformatics is a new subject that collect, store, retrieve and analyze various biological information. It's an interdisciplinary science developed by the interaction of modern biology, informatics, computer science, mathematics, statistics, physics and chemistry.The sdudy object of this article is protein structure database. The protein datas get from experiments are usually stored in databases and classified by different characters. Because during evolution protein structures are much more conserved than sequences and functions, proteins are usually classified first by their structural similarity and then by the similarity of their sequences or by the similarity of their functions. Protein structural classification may include different levels, for example, fold, topology, family, super-family, structure domain, secondary structure and so on. There have been many protein classification databases which have appeared on the World Wide Web. The three most-represented and most widely used structural classification databases are SCOP, CATH, FSSP. After analyzing the inadequate in the classification databases at present, we use multiple sequence alignment to replacce double sequence alignment and manul method to classify the proteins which have same structure. Multiple Sequence alignment can get many characters about protein, we use the method of attribute reduction in the rough set theory to select,we found the strict core length have remarkable relation with the correct protein classification So the strict core is a suitable structure character to construct the classification distance. At last, we use thirty-five family which have integrity information to test and verify, we get a good result.