The HA1 domain of influenza A viruses is critical for immune escape and host receptor recognition. However, the association between HA1 evolution and host species-specific adaptations remains unclear. We have identified the amino acid sites that are related to the host species-specific adaptations by systematically comparing the evolutionary dynamics of HA1 between human and avian hosts for both H1N1 and H3N2 viruses. We can thus refine the positively selected sites in the HA1 domain of these two virus subtypes to those that occur only in humans. We further demonstrate that the refined sets can accurately reflect influenza epidemics. Our retrospective study shows that when substitutions occur at more than three out of these identified sites in the same year, an epidemic is very likely to occur. Therefore, our approach could potentially be used in determining whether new influenza vaccines are needed.
