We attempted to understand patterns in Influenza A evolution by understanding the growth of correlated mutation graphs over a gradually increasing time period. We performed this experiment for Human H1N1 and Human H3N2 sub-systems.
Evolution of Human H1N1 Network
To visualize how H1N1 viruses in Humans evolve over time, we examined the changes in network topology over time by analyzing four different datasets listed in the following table. We started with a dataset of 698 unique strains belonging to years till 2007. For our second dataset, we created a dataset consisting of 1035 unique strains belonging to flu seasons till 2010/2011 with a maximum of 300 strains per year. Our third dataset contained strains till 2014/15 flu season and fourth dataset comprised of unique strains till 2016/17 flu season.
| NAME | #STRAINS | COMMENTS |
| HUMAN_H1N1_TILL_2007 | 698 | 698 unique strains of H1N1 belonging to years till 2007 |
| HUMAN_H1N1_TILL_10_11 | 1035 | 1035 unique strains of H1N1 belonging to flu seasons before 2010/11, maximum 300 per year |
| HUMAN_H1N1_TILL_14_15 | 1448 | 1448 unique strains of H1N1 belonging to flu seasons before 2014/15, maximum 300 per year |
| HUMAN_H1N1_TILL_16_17 | 1769 | 1769 unique strains of H1N1 belonging to flu seasons before 2016/17, maximum 300 per year |
Node and edge counts for these datasets are depicted in figures below.
These results suggest that there is a sharp increase in epistatic changes in Influenza genome between 2007 and 2010/2011 flu season during which a pandemic H1N1 occurred. There is a gradual normalization in these changes in subsequent years where the genome continued to see lesser covariance in mutations. Both the node and edge count observations confirm this hypothesis. If the trend continues (where we do not observe any new H1N1 pandemics soon), we should continue to see a similar pattern with the long-tail in the number of nodes and a gradual decrease in number of edges in the coming years.
To visualize how H3N2 viruses in Humans evolve over time, we examined the changes in network topology over time by analyzing four different datasets over an increasing 10-year window (listed in table below).
| NAME | #STRAINS | COMMENTS |
| H3N2_HUMAN_TILL_2016 | 1561 | Unique – HUMAN H3N2 – BEFORE 2007 – MAX 300 per year |
| H3N2_HUMAN_TILL_2010 | 1913 | Unique – HUMAN H3N2 – BEFORE 2010 – MAX 200 per year |
| H3N2_HUMAN_TILL_2014 | 2154 | Unique – HUMAN H3N2 – BEFORE 2014 – MAX 150 per year |
| H3N1_HUMAN_ALL | 1940 | Unique – HUMAN H3N2 – ALL – MAX 100 per year |
Compared to the H1N1 evolution plots, the H3N2 plots show a far lesser degree of change over the last 10 years suggesting a more stable, less dynamic genome. Also, the number of nodes and edges in the H3N2 network are substantially lesser than the number of nodes and edges in H1N1 network.
