ABSTRACT
Detecting significant dN/dS ratios in molecular sequence evolution is perhaps one of the best examples of natural selection in action. However, considerations include taking into account the multiple hit phenomenon, by use of the Jukes–Cantor correction, and incorporating models into the estimation of dN and dS. Different parts of proteins have different functions, and hence the selection on these domains within a single protein might differ. Indeed, selection at disparate sites might vary across a protein. Methods for detecting the degree of natural selection at individual sites in a molecule have been developed to make the estimation process more accurate. Evolutionary forces might change over time. This factor influences the degree of selection at different times in the divergence of groups of organisms. Methods for partitioning the impact of selection across phylogenetic branches have therefore been developed to tease apart the influence of natural selection on molecular sequences. Maximum likelihood approaches are implemented in selection studies at the molecular level and allow one to take advantage of phylogenomic information to make the estimations of dN and dS. ML approaches also permit simultaneous estimation of parameters, correction for multiple hits and other sequence change anomalies, and evaluation of evolutionary pathways between codons.
