Navigating Molecular Evolution Using Substitution Mappings

Since first proposed in 2002, Bayesian substitution mappings have found less use than might have been expected in the field of molecular evolution. Here, I first create a sequence simulator capable of generating true substitution mappings for simulated data under any time-reversible model in order to facilitate study of substitution mappings. I then investigate the utility of substitution mappings for two applications: detection of coevolving residues and scoring of amino acid substitution radicality to test the predictions of the nearly neutral theory. I find that mappings perform poorly for coevolution detection, but using mappings to find the radicality of an average amino acid by scoring each observed substitution works well. Overall, substitution mappings look to be a potentially useful tool for some types of molecular evolution studies.