The Early Warning Model For Influenza A Outbreak Based On Protein Feature Information

Influenza A virus is prone to mutations and susceptible to human beings and spread in the crowds when affected by the external environment or other factors. Therefore there is one conventional outbreak every few years and one large scale outbreak every ten years or so. Large scale outbreak of influenza A totally occurred four times from the beginning of 20 th century, in which the outbreak of the Spanish influenza in 1918 was the most extensive and influential. There are about 1 billion people worldwide who were infected by influenza, in which 40 million of them were killed. Therefore, the research on early warning of pandemic influenza outbreak is significant and the bioinformatics method is a kind of very good research method. In this paper, the influenza A virus protein sequences are used as the object of study to construct the mutation model, the protein dynamic network of influenza A(H1N1) virus is constructed by using the feature information among proteins to provide an effective early warning signal for the outbreak of influenza A. The main work is as follows:(1) In this paper, we select the HA protein sequence of pandemic virus reported from 1916 to 2014, with the exception of missing data for individual years. Firstly, we make comparison of the amino acid sequences of HA protein in s-1 year and pick up the amino acids with the highest number of occurrences and sequentially form a new sequence. Secondly, the standard deviation of the annual HA protein sequence is calculated by comparing the sequence formed with the HA protein sequence appeared in s year. Finally, the annual HA protein mutation is calculated accordingly. When the mutation value is relatively large, the changes of influenza A virus HA protein in s year are relatively larger than those in s-1 year. That is to say the mutation rate is relatively high. Therefore, by observing the mutation value changes, you can intuitively discover the conditions of the influenza A virus HA protein mutation.(2) Based on the appeal method, we can calculate the standard deviation, arithmetic mean and mutation variation of influenza A virus protein every year by selecting the amino acid sequences of ten sorts of influenza A virus proteins from 1933 to 2015, with the exception of missing protein data for a few years. Twenty-three feature information of ten kinds of protein sequences are constructed by calculating the frequencies of 20 kinds of amino acids in influenza A virus proteins separately. Ten core proteins of protein dynamic network biomarkers are defined each year. Feature Distance between core proteins as well as Feature Distance between Non-core proteins and core proteins are respectively calculated in this paper. And then a pandemic virus network is built by the usage of correlation between biomarkers. Application of the dynamic showed in this network as well as combination with the nature of influenza A virus protein dynamic network biomarkers, a composite index ? is obtained. The composite index ? can provide reliable and significant information to predict the critical outbreak or outbreak of influenza A virus, which indicates that we can determine the status of the influenza A virus more stably and accurately by considering the dynamic network biomarkers on the network level of the influenza A virus.By constructing the mutation model, the protein dynamic network of influenza A virus is constructed by using the feature information between proteins, which provides an effective early-warning signal for the possibility of the outbreak of influenza A. This study is of great significance for the early warning of influenza A virus.