A New Representation Method of H1N1 Influenza Virus and Its Application

摘要

Based on the 38,899 pieces of H1N1 virus protein sequences from 1902 to 2013 in the world, the 1805 H1N1 virus sequences with HA and NA protein are selected according to viruses occurred at the same time and place. A new representation of feature vector for protein sequences is proposed by the physicochemical properties of amino acids and coarse graining theories. The 20 kinds of amino acids are divided into 4 classes and connected with each other to construct 16-dimensional feature vectors to represent HA and NA protein sequence, respectively. The whole protein sequence is represented by a 32-dimensional feature vector, which combines the feature vectors of HA and NA protein sequences, and the optimal cluster of the H1N1 influenza virus is obtained by the structural clustering. The relationship between HA and NA protein structures and the outbreak of H1N1 virus protein sequences is analyzed by selecting the representative elements and constructing evolutionary tree. The results show that the new representation of feature vector for protein sequences is reasonable, and large amount of data confirms that HA and NA protein sequences play a direct and important role in the outbreak of H1N1 influenza virus.