The Research And Implementat On Algorithm And System Of Protein’s Spectral Clustering Analysis Based On Noise-Reduction

Bioinformatics is a emerging interdisciplinarity, containing the modern life science and information science、math、statistics、physics and chemical and so on.It utilizes the method of computer technology and information theory to gather、store、 deliver, retrieve、analysis and explicit the protein、nucleic acid sequence and other biological information, meanwhile it can help comprehend biology and genetic information. Nowadays the main research content of bioinformatics still is the construction of sequence map, the analysis and identify of the new gene, it has became the indispensable research target in Genomics and Proteome[1]. So the rapid development of bioinformatics not only has profound influences on life sciences, but also promote the progress of the area of life sciences and others.In the process of protein evolution, the amino acid sites are not independent of each other, but they have some interaction pattern[2][3][4]. Some amino acid sites may stay far away in the one-dimensional spatial structure, but they may contain relevance and can also form the different protein sector. These structure elements stay relatively independent in the function and evolution, but exit obvious interdependency in the inner of the structural element. Therefore researching the co-variation of protein’s internal structure is beneficial for the details of protein function research. In this article, we propose a new algorithm BIFANR(Bi-factor Analysis Based on Noise-reduction) for detecting protein sectors in amino acid sequence. Subsequently for the former protein sectors, we should inspect the internal interdependency, statistical independence, evolution rate analysis, evolution independence analysis and so on, while the final result represents that each of the amino acid sequence of the internal protein sectors has close connection, meanwhile each of protein sectors has obvious independence. Othermore, the experiment result also reveal that each of protein sectors evolves in different directions. Compared with other algorithms, this article’s algorithm possess the high accuracy and robustness, it doesn’t easily be influenced by the noise sites, separating large number of noise sites from nonrandom protein sectors.In a word, if we need transform the bioinformatics problems to the handling problem of the numerical symbols, we must develop the computer’s skill of information handling to develop the new analysis theory, method, technology, tools. We can carry out effective communication with international and domestic system, and establish broad, close contact with biology research method and platform technology. Therefore we integrate our algorithm into this article’s system with popular Internet technology, The Internet technology is not only the main method of producing bioinformatics data, but also is the critical means of validating bioinformatics research result. This article’s online service system, that is BIFANR, offers the communication platform based on this algorithm, it utilizes Java software calling MATLAB, new Web technology JSP and some components to achieve this system’s general design and implementation. According to the format of fasta or fas files uploaded by users, this system can help users to detect coevolving Amino Acid Sites in protein, then display the three-dimensional structure of protein sectors and amino acid sites produced by experiments. Our research bring great break through to Proteome.