Assessing Uncertainity on Phylogenetic Tree

Abstract

Maximum likelihood inferred topologies are commonly used to draw conclusions in evolutionary biology and molecular evolution. These computations are based amongst others on the observed nucleotides and as such it is subject to sampling error. In this talk we critically review classical measures of the robustness of the inferred tree. In particular, we present some simulations to clarify pro and cons of classical bootstrap approach. Then we extend resamling tools to characterize influential sites. Finally, using concentration measure tools, we also propose an original method to bound the variations of the computed likelihood around its true value and the probability that a phylogeny has a better likelihood than another one “just by chance”.　　　 This is a joint work with Mahendra Mariadassou.