Confidence measures for evolutionary trees: applications to molecular epidemiology

Recently there has been increased use of statistical hypothesis testing in settings where the phylogenetic (evolutionary) relationship among organisms (taxa) should be considered. For example, we have a current need to assess our confidence in assigning influenza sequences to the time period or species. We do so by using both: pattern recognition methods that do not explicitly account for the evolutionary relationships; and phylogenetic tree estimation methods that do explicitly estimate the evolutionary relationships (branching order) among the specified taxa. We compare our conclusions under both approaches for sequences from the Hemagglutinin gene of the human influenza RNA virus that are evenly distributed around the world from 1985 to 1996 and for sequences of the Nucleoprotein gene of the influenza RNA virus from three host species.