Figure 4

Interactive hierarchical aggregation for amino acid sequences. Using iHAT to find sequence positions correlated with virulence in 15 sequences of the neuraminidase protein of H5N1 influenza virus samples. (A) the unfiltered alignment using the color map based on Venn diagram grouping; (B) after removing uninformative columns (parameters shown in the dialog window below the alignment), only ten positions remain. (C) sequences were sorted by virulence and internal nodes were added aggregating by 2 levels of virulence; (D) after aggregation on (collapsing) the internal tree nodes, the final alignment of the two aggregated sequences (labeled with unique numerical identifiers starting with 'A') clearly shows positions correlated with virulence.