Probabilistic Models for the Study of Protein Evolution

Abstract

Model choice is one of the central statistical issues in the study of protein evolution. Models are indispensable tools for characterizing the process of protein evolution, many aspects of which are not easily amenable to direct experimentation. In these cases, only probabilistic models can assess the fit between assumptions and data. Ideally, a probabilistic model of protein evolution would provide a good statistical fit to the data and would simultaneously be parameterized in a manner that facilitates biological insight. In addition, probabilistic models of protein sequence evolution can provide the foundation for likelihood-based methods of phylogeny reconstruction and protein structure prediction. In this chapter, models of protein evolution are reviewed. The strengths and limitations of these models are emphasized. Keywords: amino acid replacement; evolutionary model; phylogeny; probabilistic model; protein evolution; protein structure