As shown by the 2009 Swine Flu outbreak, the influenza epidemics are often caused by human-adapted influenza viruses originally infecting other animals. The influenza viruses infect host cells through the specific interaction between the viral hemagglutinin protein and the sugar molecules attached to the host cell membrane (called glycans). The molecular mechanism of the host switch for Avian influenza viruses was thus believed to be related to the mutations that occurred in the viral hemagglutinin protein that changed its binding specificity from avian-specific glyans to human-specific glycans. This theory, however, is not fully consistent with the epidemic observations of several influenza strains. I will introduce the bioinformatics approaches to the analysis of glycan array experiments that revealed the glycan structural pattern recognized by the hemagglutinin from viruses with different host specificities. The glycan motif finding algorithm adopted here is an extension of the commonly used protein/DNA sequence motif finding algorithms, which works for the trees (representing glycan structures) rather than strings (as protein or DNA sequences).