Prediction of amino acids determinants of trends in viruses and application to the H1N1 and H5N1 influenza virus
Mutations in specific amino acid positions may lead to changes in the trend of viruses, e.g., accelerated transmission rate or pathogenicity. I will present a supervised machine learning approach for predicting such amino acid determinants based on the protein sequences and their known segregation with respect to the trend. The method will be used to analyze proteins of the H1N1 and H5N1 flu viruse.
The outbreak of the swine-origin pandemic influenza virus H1N1 (pH1N1) in 2009 has raised substantial concerns. Fortunately, the mortality rate of this pandemic has been relatively mild, but pH1N1demonstrated a more efficient infection and transmission than the seasonal strain. My colleagues and I analyzed hemagglutinin, the major protein of the virus envelope, and searched for residues that differentiate between pH1N1 isolates and isolates from other H1N1 groups, namely swine strains and the common-seasonal human strains. Our analysis showed that all the residues detected as discriminating between the pH1N1 and the human circulating H1N1 hemagglutinin sequences were located in or near the known H1N1 antigenic sites. Mutations in these positions may alter the stereochemistry of the virusís antibody binding sites, thus hindering immune system recognition of the viral surface. We also detected residues differentiating between the pH1N1 and classical swine strains. Nearly all these positions were situated in or around the receptor- binding pocket. Mutations in these positions may alter binding to the host cell receptor and therefore affect the infection efficiency. The identification of these sites may prove useful for designing novel neutralizing antibodies or antiviral drugs that could inhibit viral entry. Some of our predictions were validated in experiments.
The H5N1, avian, flu is considered particularly dangerous to the human population because of its exceptionally high virulence. This is a more challenging case since it was not fixated in the human population. I will present our preliminary effort to detect specificity-determining positions in this strain.
Reference: Meroz et al (2011) Putative amino acid determinants of the emergence of the 2009 influenza A (H1N1) virus in the human population. PNAS 108:13522-13527.