Flexible protein alignment of 3D-structures allowing dynamic transformation (バイオ情報学)

概要

論文の詳細を見る
In these days attention is focused on the functions of proteins and it is known that when the 3D-structures of proteins are similar the functions of them are often similar. Thus to find the similarity of protein structures is a very important problem. However, most parts of proteins have been treated as rigid ones in the previous papers, though the protein knocks against various molecules in the solvent and is transformed. In this paper, we focus on the local structures of proteins, consist of a sequence of 4 coadjacent amino acids. We investigated the features of Local Structures on two similar proteins, and on two proteins that have no relationship, Then by using the features and the probabilistic argument, we defined a new Similarity Score (SS) between two proteins that can be calculated in the time proportional to the size of the proteins. We show that we can get almost the same results in much shorter time compared to the results that uses RMSDh. We also show SS can be used to point out the existence of hinge regions with high probability. Our method is several hundreds times faster than existing methods and by the fact we can make some groups of similar proteins allowing some hinge regions, or to reduce the candidates of similar protein pairs before using some slow algorithms.
一般社団法人情報処理学会の論文
2008-03-03