Methodologies for Phylogenetic Inference

Abstract

Abstract Phylogenetic inference from homologous molecular sequences is key to hypothesis testing and problem solving not only in evolutionary biology but also in a wide variety of other fields – from medicine to ecology. Model‐based phylogenetic methods rely on Markov substitution models to describe the molecular evolution as a stochastic process of character substitution over time on a phylogenetic tree relating the sequences. Model parameters are estimated with standard statistical inference methods, namely, Bayesian and maximum likelihood approaches. A typical phylogenetic analysis first infers a multiple sequence alignment. Given this alignment, a phylogenetic tree is then estimated together with branch lengths and model parameters. Ideally, the alignment and phylogeny should be estimated simultaneously, amongst others to take alignment uncertainty into account. Key Concepts Sequence Alignment estimates an assignment of homologous molecular characters, that is, nucleotides, amino acids or codons related by common ancestry. Phylogenetic Tree is the hierarchical representation of evolutionary relationships between homologous molecular sequences. The leaves of the tree usually represent the present day sequences, while the internal nodes represent the common ancestors. Model of Molecular Evolution is the mathematical description of the process of sequence change through time, such as character substitutions, insertions and deletions. Phylogenetic Likelihood is the probability function of observing the sequence data given the model of molecular evolution and phylogenetic tree. Frequentist Phylogenetic Inference relies on optimised phylogenetic likelihood to estimate parameters of a model of molecular evolution and a phylogenetic tree. Bayesian Phylogenetic Inference relies on phylogenetic likelihood and a prior probability distribution of parameters to obtain a posterior probability distribution of parameters of a model of molecular evolution and a phylogenetic tree. Branch Support quantifies the uncertainty of phylogenetic inference by assigning statistical confidence to the inferred partitions (i.e. clades) on a phylogenetic tree. Alignment Uncertainty quantifies the statistical confidence of sequence alignment, which compounds the uncertainty of phylogenetic inference.