The importance of using realistic evolutionary models for retrodicting proteomes

Abstract

The reconstruction of phylogenetic trees from molecular data requires selecting models of molecular evolution that adequately describe known processes of change. Operationally, these models optimize molecular changes along branches of the trees. The underlying processes must be realistic and must comply with well-supported biological assumptions. In a recent paper, a new model of proteome evolution that penalizes growth of the protein world provides an 'upside down' phylogeny and identifies a very complex ancestor of diversified life. Here we show that the model is phylogenetically self-inconsistent and at odds with considerable background knowledge, including the scale-free property of domain networks, genomic scaling laws, and the principle of continuity that supports the tenets of ideographic analysis and evolutionary thinking. While technical and conceptual limitations invalidate the main conclusions of the study, including the existence of bottlenecks in protein evolution caused by planetary cataclysms, we use the example to highlight the complexities and pitfalls of retrodiction in phylogenetic and phylogenomic analyses and reexamine the framework of ideographic exploration that is used in scientific inquiry.