Preprint Open Access
A principal component analysis of a multiple sequence alignement of hemagglutinin sequences of subtype H1 has been performed, the sequences being encoded using the amino-acid property that maximizes the weight of the major component. In the case of this alignment, it happens to be a well-known hydrophobicity scale. Interestingly, sequences coming from human have large positive amplitudes along the major component before 2009, and large negative ones afterwards. This strongly suggests that the 2009 pandemic was associated to a major change in the hydrophobicity pattern of hemagglutinin.
The present analysis also highlights the high variability of viral sequences coming from swine. At a more general level, the method proposed herein allows to describe a sequence coming from an alignment with a set of numbers, the original point being that the choice of the corresponding property is driven by the data.