A longstanding difficulty in phylogenetic inference entails selection of appropriate analytic methods for a given lineage and identifying suitable models of sequence evolution, with associated parameters and parameter values. Errors in selection of a suitable algorithm or evolutionary model are known to bias phylogenetic analyses, yielding unreliable estimates of evolutionary relationships. Our research project aims to design a program that leverages the statistically predictable behavior of a set of evolutionary mechanisms to generate datasets—via Monte Carlo simulation—that faithfully retain salient genomic features of evolving clades in the real world. Our project seeks to develop an application, implemented in the Python programming language, which will simulate genomic evolution of a single ancestral sequence into a lineage of progeny sequences. The program will generate a set of terminal sequences and a known topology that will subsequently be supplied as input to a range of competing phylogenetic algorithms. The resulting performance metrics will provide insights into the efficacy of commonly used phylogenetic algorithms and software packages. These insights, in turn, will allow researchers to delineate the strengths and weaknesses of competing approaches when applied to datasets with particular dynamic and compositional properties. Hence, this project has the potential to address an urgent unmet need for informed algorithm and model selection during phylogenetic analysis.
