Hierarchical Clustering Using Constraints

Abstract

Abstract. We describe a new supertree algorithm that extends the type of information that can be used for phylogenetic inference. Its input is a set of constraints that expresses either the hierarchical relationships in a family of given phylogenies, or/and other relations between clusters of sets of species. The output of the algorithm is a multifurcating rooted supertree which satisfies all constraints. Moreover, if there were contradictions in the set of constraints the corresponding part of the supertree is identified and its set of constraints is displayed such as the user may decide to modify or keep it. Our algorithm is not affected by the order in which the input phylogenies or other constraints are presented. We apply our method to a number of data sets.