Very large phylogenies are being built today in order to study virus evolution, trace the origin of epidemics, establish the mode of transmission, and survey the appearance of drug resistance. However, no tool is available to quickly inspect these phylogenies and combine them with extrinsic traits (e.g. geographic location, risk group, presence of a given resistance mutation), seeking to extract strain groups of specific interest or requiring surveillance.

We propose a new method for obtaining such groups, which we call phylotypes, from a phylogeny having taxa (strains) annotated with extrinsic traits. Phylotypes are subsets of taxa with close phylogenetic relationships and common trait values. The method combines ancestral trait reconstruction using parsimony, with combinatorial and numerical criteria measuring tree shape characteristics and the diversity and separation of the potential phylotypes. A shuffling procedure is used to assess the statistical significance of phylotypes. All algorithms have linear time complexity. This results in very low computing times.

There is three ways to compute a PhyloType analysis. The first one is the use of the web interface (see the "Interface" link above), the second one is to use a standalone module to download and the third way is to use a "batchmode" module (for massive PhyloType analyses, without a graphical user interface).