A Value-Added Database for Efficient Mining of Influenza Virus Genetic Markers

Influenza viruses are known to infect various hosts. In particular that influenza A viruses (IAVs) can infect a wide range of avian species and mammals. Although IAVs in avian are generally not transmissible to human, recent outbreaks indicated that some high-titer or mutated avian IAVs resulted in severe human cases, some were fatal. Because of very different receptor-binding mechanisms between avian and mammalian cells, it has been hypothesized that a crossing-species boundary “separates” the avian IAVs from becoming transmissible to human. Specific mutations occurring on those points may enable an avian influenza virus to become a human virus. In this study we implemented a plateau-searching algorithm in locating one entropy threshold, based on which the species-associated signature positions between avian and human influenza A virus protein sequences can be reported. The searching result is found to include all previously located signature positions with utilizing an entropy threshold solely computed from the amino acid composition of a well-known species-associated virulence marker PB2 627. Our method not only is able to discover more signature positions, customized entropy threshold can be found from each protein based on participated protein sequences. We believe the successful locating of such plateau is a result of natural aggregation of viral swarming. Similar approach in detecting such swarming can be applied to investigate other phenotypes involving highly mutated viral genomes such as influenza A viruses.

Project ID:PB10408-5760
External Project ID:MOST104-2221-E182-064