The hemagglutinin mutation E391K of pandemic 2009 influenza revisited

- Use of small sequential data sets lead to overestimating mutation rates in influenza.
- The incomplete lineage sorting and missing ancestors are the two main reasons.
- Together with a “greedy” character of the Neighbor Joining algorithm.
- optimizing codon Ist-3rd position mutations allows a better phylogenetic disentanglement.
- Leading to 3.83 × 10−4 substitution rates for HA (9.64 × 10−5 for the non-equivalent AA).