Likelihood-based clustering (LiBaC) for codon models

This chapter focuses on the likelihood-based clustering (LiBaC) method and its application to models of codon evolution. The LiBaC method for partitioning sites into groups, or 'clusters', where each group has a different model, was developed by Bao et al. (2008). LiBaC can provide reliable parameter estimates under an appropriate model when the process of evolution is very heterogeneous among groups of sites, and it can be used to identify sites subject to positive selection. LiBaC can be employed to search for genes having evolved under positive selection pressure. This chapter presents a large-scale survey of genes encoding transmembrane proteins-a hallmark of these genes is substantial evolutionary heterogeneity among sites. It also reviews the conclusions derived by Bao et al. (2008) from their simulation studies, and examines the effect of different posterior probability cutoffs on LiBaC performance.