Influenza viruses are the leading infectious viruses that have caused a constant influenza pandemic worldwide, beginning with the Spanish flu in 1918 and the Asian flu, the Hong Kong flu and the 2009 swine flu. Influenza viruses increase the likelihood of transmission and epidemics in human populations by overcoming species barriers and increasing adaptability to human hosts through a series of evolutionary events. Host-tropism determinants of influenza viruses, which cause changes in host range or increase the likelihood of interaction with a particular host, are crucial in understanding the way viruses infect and spread to heterogenous hosts across interspecies barriers. Therefore, in this research, we analyzed influenza virus protein by using bioinformatics techniques. We have searched for a machine learning algorithm and the major host determinant that can be used effectively in host classification and prediction, and derived the physicochemical properties of amino acids useful for host classification and prediction by computer programming and statistical analysis. As a result of the research, a database IVDB-HT was constructed to provide information on influenza virus host pathogenesis, biological and literature information, and epidemiological information on influenza. Bioinformatics analysis was performed based on the constructed database. As a result, among the six host tropism determinants, HA protein was the most important determinant for host classification. Also, random forest algorithm probed to be the most accurate and effective predictor for the host classification among the three machine learning algorithms – random forest, naïve bayes classification, k-nearest neighbor method. In addition, computer programming was used to select conserved amino acids and their location, which are different between hosts, and to present them as location markers which can be useful for host classification. Finally, the ANOVA and the post-hoc test showed that the physicochemical properties of the amino acids constituting the protein sequences combined with the location markers were significantly different depending on the host. As a result of this research, it is expected to improve the understanding of the physicochemical properties of amino acids as well as the molecular biologic characteristics that affect host tropism of influenza viruses. In addition, it is considered that the determinants of host tropism and location markers could provide a basis for the physicochemical properties of the host determinants to classify, determine and predict the hosts of newly or re-emerging influenza viruses. Furthermore, if the bioinformatics methods and results of this research can be used for human host tropism discrimination of new and variant influenza viruses, it will be a useful scientific basis for the prevention of human infection and propagation.
PDF
download
878987797
028-85220240
