RNA-Based Phylogenetic Methods

RNA molecules are used extensively in phylogenetic studies. Their strongly conserved secondary structure can be used to improve the alignment of highly diverged sequences, and this in turn can improve phylogenetic accuracy. The conservation of secondary structure in RNA molecules is a consequence of compensatory substitutions or covariation: when a base changes on one side of a helix this will usually be compensated for by a similar change on the opposite side, in order to maintain structural stability and biological function. The standard nucleotide substitution models for DNA, introduced in Chapter 4, ignore this covariation of base-paired sites, which leads to biased inference results and unreliable posterior probabilities of subtrees (clades). In the present chapter we discuss how these shortcomings can be overcome with a generalised substitution model that treats base-paired sites in the RNA helices, rather than single nucleotides, as a unit of evolution. We will present results from a study of mammalian evolution using Bayesian inference methods and explain why it is particularly important to include structure information when estimating Bayesian posterior probabilities.